When I first started using ChatGPT, most of my prompts were pretty basic. I’d ask for side hustle ideas or advice on preparing for a job interview, then wonder why the answers felt so generic.

Eventually, I realized ChatGPT could help me write a better prompt before answering it. The technique is known as “meta-prompting,” It's a way to give ChatGPT a chance to spot missing details, clarify what you want and structure the request in a way that produces a more useful response.

I now use one prompt ChatGPT created to improve nearly every prompt I write. Paste in your original request, answer any follow-up questions and it will rewrite the prompt around what you’re actually trying to accomplish.

“Meta-prompting” strengthens my prompts and it’ll do the same for you

As a clear example of how “meta-prompting” works, I started one of my most recent conversations with ChatGPT by simply asking, “How can I learn how to create thumbnails faster? I'm a content creator who's just starting and wants to create quality Photoshop thumbnails for my videos about video games.”

Here’s the prompt that ChatGPT gave me when I asked what sort of prompt would be best to input when I wanted my original request to make my goal more understandable:

Identify what is missing, ambiguous, or poorly framed in my original prompt. Make the minimum necessary assumptions to turn it into a high-quality expert request. Do not merely make my wording sound more sophisticated—improve the underlying question, criteria, and desired outcome.

ChatGPT took a look at my original prompt and detailed a bunch of extra bullet points that helped me figure out exactly what I was asking for. Those bits of info that I needed to specify to help the chatbot lock in on my overall mission included what “faster means, my current Photoshop skill level, what makes a “quality” thumbnail, whether you're trying to develop your own style, what I want to practice and what constraints I’m working under. Then the chatbot framed my prompt to fall more in line with the request: “How do I build the skill of making compelling gaming thumbnails quickly?”

The improved expert version of my original prompt was then delivered as follows:

I'm a beginner content creator making YouTube videos about video games, and I want to learn how to create high-quality Photoshop thumbnails much faster.

Design a practical learning system that helps me become both better at thumbnail design and faster at producing thumbnails, rather than simply teaching me Photoshop features.

Identify the most important thumbnail-design skills I should learn first, including composition, visual hierarchy, contrast, color, typography, character/game-image placement, and creating curiosity.

Identify the Photoshop skills and shortcuts that provide the biggest payoff specifically for gaming thumbnails.

Give me a 30-day practice plan that takes me from beginner to being able to create polished thumbnails efficiently.

Explain what I should practice repeatedly versus what I should learn only once.

Give me a deliberate-practice exercise for recreating successful gaming thumbnails, so I learn the underlying design principles rather than simply copying them.

Show me how to analyze a successful gaming thumbnail before I recreate it.

Explain how to build reusable assets, templates, layer structures, effects, fonts, and PSD files so each new thumbnail takes less time.

Give me a recommended workflow for going from a blank Photoshop document to a finished thumbnail, with an emphasis on speed.

Identify the most common mistakes beginners make that make gaming thumbnails look amateurish.

Explain how I can tell whether a thumbnail is actually improving without relying solely on views or click-through rate.

Assume I know very little about professional thumbnail design but am willing to practice consistently. Prioritize the 20% of skills that will produce 80% of the improvement.

My goal is not to become an expert Photoshop user. My goal is to be able to consistently create eye-catching, professional-looking gaming thumbnails in roughly 15–30 minutes.

With this new and improved prompt, I was provided with a command for ChatGPT that took my originally vague request and turned it into a more substantial prompt meant for skill-building. Attaching the meta-prompt to your original prompt will turn it into something similar to mine, which is a prompt with a clear description that locks in on a clear number of goals.

The takeaway

Meta-prompting is one of the most efficient ways to turn your simple prompts into more comprehensible ones and compel you to improve upon the way you jot down your future prompts.

Plain prompts like “Give me some career advice” and “How can I reduce stress?” just won’t do at this point if you regularly interact with ChatGPT—use the meta-prompt to turn that first prompt into a more decision-oriented framework and the second one into a plan that delivers immediate relief from defined causes of mental strain, for example.

Prompt refinement will make your ChatGPT experiences more accommodating in the long run.

More from Tom’s Guide

• Gemini 3.7 Flash is here — and this prompt proves why it's known as Google's 'workhorse'
• Meta files patent for AI facial recognition glasses that identify people around you and record automatically
• I challenged ChatGPT and Gemini to find the best ways to unwind — these 7 ideas made National Relaxation Day even better

ChatGPT now has Projects, Skills and Computer Use just like Claude. Canvas, one of my very favorite features, can render interactive React and HTML, like Claude too. Several features that once made Claude very unique are no longer exclusive to Anthropic’s chatbot.

However, there are differences that haven't disappeared and in fact, have gotten even more specific. Claude can reach into parts of my phone that ChatGPT cannot, and it gives me more control over what happens to my personal context. Its Artifacts have also moved well beyond simple code previews.

And while I still use ChatGPT, here are five things that it just can't replicate yet that still make Claude stand out.

1. Claude can work with native apps on my iPhone

ChatGPT can connect to services including Gmail and Google Calendar, but it doesn’t have the same access to the apps built into my iPhone.

Claude can prepare a message actually in Messages, draft an email in Mail, create an event in Apple Calendar, add something to Reminders or send a location to Maps. The AI assistant presents an action for me to review first, and iOS hands it to the appropriate app.

That makes the feature feel like a natural extension of a conversation. I can ask Claude to turn plans we just discussed into a calendar event without opening another app and entering everything again.

The feature works across Claude’s plans, according to Anthropic’s iOS documentationand Android users get an even broader set of controls. Claude can work with messaging, email, calendars and maps while also setting alarms and timers through supported Android apps.

ChatGPT has become far more capable at taking actions, particularly through its connected apps and Work mode, but it still doesn’t offer this kind of direct integration with Apple’s everyday system apps.

2. Claude can import what another AI remembers about me

Starting over with a new chatbot can be surprisingly annoying. When you start fresh, the assistant doesn’t know how you work, what you’re interested in or how you prefer it to answer. All of the context built up over months stays behind with the old service.

Claude now offers a built-in way to bring some of it with you. You first ask the other assistant to produce a summary of what it remembers, review that information and remove anything you don’t want to move, then paste it into Claude’s memory-import flow.

Anthropic even provides a prompt for extracting that context and lets you view or export Claude’s memory if you decide to leave later.

Although you can view and edit saved memories within ChatGPT, it doesn’t provide a comparable memory-import tool that turns context from another assistant into ChatGPT memories.

You could recreate some of this manually through custom instructions, but that isn’t the same as having a dedicated migration flow. Claude is currently making it easier to arrive with your history and leave with it again.

3. Claude’s Incognito chats can still know who I am

Sometimes I want a conversation to disappear from my history without making the assistant forget everything useful about me. Unlike other chatbots, Claude’s Incognito mode allows that separation. The conversation isn’t added to my chat history, but Claude can still use my profile information, preferences and memory while answering. I can have a temporary conversation without giving up the personalization I normally rely on.

ChatGPT’s Temporary Chat takes a different approach. OpenAI describes it as a blank slate: ChatGPT doesn’t access or create memories, although it will continue to follow any custom instructions I have enabled.

There are reasonable privacy arguments for both designs. ChatGPT’s approach provides a cleaner break from previous conversations. Claude’s is more convenient when I want a chat to remain out of my history but still want the assistant to understand my usual preferences.

4. Claude can build a dashboard that updates itself

ChatGPT Canvas and Claude Artifacts can both turn a prompt into an interactive page. Either assistant can build a calculator, visualization or small browser game and let me use it alongside the conversation.

Claude’s Live Artifacts go a step further though. In Claude Cowork, I can create a persistent dashboard that draws from connected apps and local files. When I reopen it, the Artifact can retrieve current information instead of showing the same data that was available when it was created.

That could be a project tracker built from Asana and Linear, a morning page showing calendar events and unread message or a dashboard that checks a folder containing regularly updated reports. Claude saves it in the Artifacts area, keeps a version history and lets me return to it without digging through the original chat.

Live Artifacts currently require a paid Claude plan and the desktop app. They are also different from ordinary Claude Artifacts, which remain available on the web and mobile.

ChatGPT Canvas can render an interactive interface, and ChatGPT can retrieve information from connected apps during a conversation. It does not currently turn a Canvas into a persistent dashboard that refreshes itself from those sources whenever I open it.

5. Claude can put AI inside an Artifact without an API key

This may be the most interesting difference because it changes what a non-developer can build.

Claude can create an Artifact with Claude itself built into the experience. That could be an interactive tutor, a writing coach, a quiz that adapts to each answer or a tool that analyzes information entered by the user.

Normally, publishing an AI app means setting up an API account, protecting a key, deploying a backend and paying for every request. Claude handles that infrastructure for an AI-powered Artifact. People sign in with their own Claude accounts, and the usage counts against their subscriptions rather than the creator receiving the bill.

Claude Artifacts can also use persistent storage, allowing an app to remember text-based information between sessions. Depending on how the creator configures it, that data can be private to each user or shared across everyone using the Artifact. Paid users can connect Artifacts to outside services through MCP as well.

ChatGPT can write the code for an AI app, and OpenAI gives developers tools for building one. But, OpenAI's feature doesn’t currently offer the same one-step path from a conversation to a hosted, AI-powered app that other people can immediately use without an API key.

Bottom line

ChatGPT has caught up with Claude in several areas, but it still can’t do everything on this list. It's why I regularly use multiple chatbots within my AI stack.

AI assistants can do very similar things, but their unique features make them an asset, especially when added to other AI assistants. For that reason, I’m not replacing ChatGPT with Claude. I use both every day, and each one is better suited to different jobs. But when I need one of these five features, Claude is the only one that can do it.

More from Tom's Guide

• Gemini 3.7 Flash is here — and this prompt proves why it's known as Google's 'workhorse'
• 5 Gemini features ChatGPT still can’t match
• Meta files patent for AI facial recognition glasses that identify people around you and record automatically

Maneuvering through life means you’re bound to come up against some major dilemmas that require you to make a tough decision.

Typically, one of those decisions can keep you in a comfortable position, while the other could potentially be transformative and simultaneously break you out of your comfort zone. I can certainly relate to that sort of situation, since I made the tough decision to attend college out of state instead of staying in my home state to pursue my degree. Forcing myself to examine the choices that cater to my safety and the ones that could maximize my upside while pushing me to do something outside of my wheelhouse is trying but necessary.

There’s one conundrum that has plagued me for the longest time now. And while I’ve gotten plenty of feedback on how to remedy that issue from loved ones, I’m usually left with more questions regarding what I should do to remedy it.

For my latest “help me fix my problems” session with ChatGPT, I used the “strategic discomfort” prompt to tackle said issue. It rose to the occasion to be as real as it gets and proved to be tough yet fair in its analysis.

Uncomfortable truths backed by uncomfortable decisions

The strategic discomfort prompt is as follows: You're my strategic coach. Your goal isn't to motivate me—it's to identify where I'm playing too safely. Show me the uncomfortable decisions that would create the biggest upside.

As for my longstanding problem, I asked for its advice on the sort of decisions I should make to help me remedy the overwhelming feeling of being afraid of disappointing people. I’ve always felt like this issue puts the wants and needs of others ahead of mine and leads me to take on all the tasks that people ask of me since I’m always keen to please.

ChatGPT’s opening line showcased its transformative state from an average AI assistant that produces nothing but flattering advice to a tough coach that presents hard truths combined with viable solutions:

The uncomfortable truth: if you're afraid of disappointing people, you may be optimizing your life around avoiding other people's short-term discomfort rather than maximizing your own long-term upside. That creates a very specific strategic problem: you become reliable, agreeable, and easy to work with—but potentially under-leveraged.

Afterward, ChatGPT provided me with a list of uncomfortable decisions I should put into action that could change my way of thinking and make me approach my interactions with people in a completely different manner.

For instance, the chatbot told me to stop making myself indispensable in low-value ways.

While it stated how being the person who always responds quickly, fixes things, takes the extra assignment, etc., feels valuable, being useful isn’t the same as being strategically valuable. The uncomfortable decision the chatbot told me to consider is becoming less available for low-leverage work, so I have the capacity for high-leverage work. Which, by my definition, means ignoring requests that are routine and aren’t truly beneficial while saving my energy for requests that produce a major impact for me and anyone else involved with them.

Some of the other fixes and uncomfortable decisions ChatGPT presented to me as the best ways to confront my problem included:

• Stop saying yes before you've decided whether you actually want to: Start declining things that are merely reasonable rather than genuinely valuable.
• Let someone think you're difficult: Choose the option that advances your goals even when you know someone will be mildly unhappy with you. Not cruelly. Not recklessly. Just without treating their disappointment as evidence that you made the wrong decision.
• Ask for more before you have proof you deserve it: Make the ask while there's still a meaningful possibility of rejection. The rejection is often cheaper than the years spent quietly under-asking.
• Have the conversation you've been avoiding: Say the thing clearly rather than continuing to engineer around it.
• Choose your ambition over your current identity: Allow people to update their perception of you.
• Stop confusing disappointment with harm: This distinction could radically change your decision-making. Someone being disappointed doesn't necessarily mean you hurt them. Someone being frustrated doesn't necessarily mean you behaved badly. Someone disagreeing with you doesn't necessarily mean you made a bad decision. Someone needing to adjust their expectations doesn't necessarily mean you failed them. Try this filter: Did I violate a legitimate commitment—or did I simply fail to satisfy someone's preference? Those are very different things.
• Take the opportunity that creates asymmetric upside: When you're afraid of disappointing people, you tend to favor reversible, socially safe decisions. I'd instead look for opportunities where: Downside = uncomfortable conversation/temporary disappointment. Upside = materially better career, income, freedom, relationships, reputation, or trajectory. Those are precisely the bets you should be more willing to take. The mistake would be avoiding a 20% chance of awkwardness that comes with an 80% chance of substantially improving your situation.

Final thoughts

Getting the most realistic feedback possible from ChatGPT about my longstanding issue of being fearful of letting everyone down proved to be an eye-opening experience.

With the right prompt, it confronted my life obstacle, produced strict yet equitable advice and steered me towards changing my habits for the better. Now I’m even more inclined to experiment with the strategic discomfort prompt when I need some decision-making tips about my career, relationships, a change in environment and more.

Talking with friends, family and experts before I make life-changing decisions is a must. But I’ll also be sure to consider ChatGPT’s discomforting yet helpful counsel from now on.

More from Tom’s Guide

• Google Maps is better than ever — here are 7 'Ask Maps' prompts to try next
• I connected Claude to my daily apps — here are the 5 that actually made it useful
• This new AI assistant makes AI easier for people 65+ — no app or prompts required

Over the past few years we've seen ChatGPT and Gemini get some really useful features. Each AI assistant offers unique abilities and both ChatGPT and Gemini can handle many of the same jobs. But their differences become clearer when you move past the question and answer features.

Google has connected Gemini to services and devices it already controls, including Android and Google Photos. It has given Gemini a few creative tools that ChatGPT still doesn’t offer, too. You can create a finished song, turn a research report into a podcast and build a reusable AI version of yourself.

I use both assistants regularly, and these are five things Gemini can currently do that ChatGPT cannot do natively. Of course, availability of these Gemini features depends on your device, account, subscription and location, so you may not see every feature yet.

1. Create an original song you can download

ChatGPT can write lyrics, suggest chord progressions and help plan a song. But it still does not have the ability to take that prompt and hand you a finished track with music and vocals.

Gemini can because it offers Lyria 3 music generator directly into the app. You can describe the song you want or upload a photo or video to use as inspiration. You can even upload text messages and ask it to create a song from those. The possibilities are endless right in the chatbot as Gemini then generates the music and lyrics together.

A prompt could be as simple as: “Create an upbeat pop song about surviving the first week of school as a tired parent.”

You can download the finished result as an MP3 or save it as an MP4 with cover art. According to Google’s instructions, generating a full track requires switching Gemini to its Pro model. This is one of the clearest gaps between the two assistants. ChatGPT can help you plan the song. Gemini can make something you can actually play.

2. Search through your Google Photos without uploading anything

Finding one particular picture in a library containing thousands of nearly identical family photos can take forever. But Gemini can search a connected Google Photos account using the people, places and objects that appear in your images. You could ask it to find photos from a particular birthday, show you pictures from last summer or look for images containing a specific person.

It can answer questions using details from those photos, too. Google suggests prompts asking Gemini to identify the themes used at someone’s previous birthday parties or recall what you ate during a vacation.

ChatGPT can analyze any photo you upload, but it cannot natively search through your entire Google Photos library. You have to find the relevant pictures and add them to the conversation yourself.

There are several restrictions. Google Photos access in Gemini requires a personal Google account, isn’t available to users under 18 and only works in supported regions. You must turn on Gemini’s Keep Activity setting and certain Google Photos features, including Face Groups.

I’d think carefully before asking an AI assistant to retrieve something as sensitive as a driver’s license or medical document from my photos. For finding 47 pictures of the same child wearing their favorite blue shirt, though, it could save a lot of scrolling. If you know, you know.

3. Turn a document into a podcast-style conversation

Gemini can take a document, presentation, Gemini Notebook (formally NotebookLM) or Deep Research report and turn it into an Audio Overview.

Instead of reading a long report, you can listen to two AI hosts discuss the information in a podcast-like conversation. It’s especially useful for catching up while driving, walking or working around the house.

ChatGPT Voice can read answers aloud and discuss an uploaded document with you. It does not currently have a dedicated feature that turns the source into this kind of produced, shareable conversation.

Gemini usually takes a few minutes to create an Audio Overview. Once it’s finished, you can listen inside the Gemini app, download it or share a link. Google explains how Audio Overviews work here.

This could be useful for more than research papers. I can see myself using it to get through a long school document, meeting notes or a report I need to understand before starting work.

4. Create Personal Gemini Avatars (AI Digital Twin)

Google allows paid subscribers to build a Personal Avatar, which is a digital clone that looks and sounds like you, powered by Gemini Omni and Google Veo. You can build it right in the Gemini app.

Tap on your profile/settings, and select Personal Avatar (or tap Add Files + Avatar in the prompt window). From there, follow the on-screen prompts on your smartphone to record your face and voice.

You will be asked to turn your head side-to-side and read out lines of text so Gemini can map your facial structure and voice profile.

Once recorded, you can summon yourself in Gemini prompts using @me or @[your username].

You can also create a custom gem with a prompt such as:
"Generate a 3D digital illustration of a friendly male developer with short brown hair and round glasses, wearing a dark blue hoodie, vibrant neon-lit background, studio lighting, high resolution profile avatar."

5. Analyze a YouTube video from a link

I often find a useful YouTube video and immediately wish I could skip the introduction, sponsorship and repeated explanations. With Gemini, I can paste the link into a chat and ask for the information I actually need.

Gemini can summarize a public YouTube video, answer questions about what was discussed and pull out specific tips or instructions. I could give it a 30-minute product tutorial and ask, “What settings does the presenter recommend changing?” or “List the steps for fixing the problem, but leave out the promotional sections.”

Gemini can also find relevant videos, playlists and channels before answering questions about their content. Connecting YouTube lets it personalize some results using your viewing history.

ChatGPT can now do something similar through its Chrome extension, which uses a video’s timestamped transcript when captions are available. The difference is where it works. Gemini handles YouTube links inside its regular web and mobile apps, while ChatGPT’s officially documented version requires opening the video through Chrome.

Bottom line

ChatGPT remains my choice for plenty of everyday work, but Google is using its existing products to give Gemini abilities that are difficult for another chatbot to copy.

Google Photos search and Android control come from Gemini’s position inside Google’s ecosystem. Music generation, Audio Overviews and reusable avatars are more surprising because they change what you can create without leaving the chat. Give them a try and let me know what you think in the comments.

More from Tom's Guide

• Gemini 3.7 Flash is here — and this prompt proves why it's known as Google's 'workhorse'
• 5 custom Gemini Gems that I swear save me hours each week
• This new AI assistant makes AI easier for people 65+ — no app or prompts required

At this point, I’m nervous enough not to crack open a box of Teddy Grahams because it might not be safe to eat.

Every time I turn on the news, hop on social media or tune into my favorite morning radio shows, I see more news about a food recall. I’ve been lucky enough to avoid getting explosive diarrhea for the past few summer months after the first big recall, but now it seems like there are other unsafe foods making the rounds.

I wanted to get a primer on the most recent food recalls, learn about the best methods to avoid them & other types of contaminated food and get quick tips on how to remain safe.

To see this mission to a successful completion, I asked ChatGPT to assist me. In the end, it gave me all the helpful information I needed that I’m sure will prove equally useful for you.

The prompt that “fed” me the safety information we all need to know

I wanted to go into my latest health-centered chat with ChatGPT with a prompt that covered all of my requests. I asked the chatbot to refine my simple prompt of “Tell me about the latest food recalls and give me advice on how to remain safe from them.”

Here’s the super descriptive prompt it gave me to produce even better results:

Act as a food-safety researcher and help me identify the most recent food recalls in the United States.

Search the web for recalls announced or updated within the last 30 days, prioritizing authoritative sources such as the FDA, USDA FSIS, CDC, and official manufacturer/retailer notices. Do not rely on social media or unverified reports.

For each significant recall, tell me:

Product name and brand

Reason for the recall

Contaminant or hazard involved, if known

Product codes, lot numbers, UPCs, expiration/use-by dates, or other identifying information

Where the product was sold

Recall date and whether the recall is still active

What consumers should do if they have the product

Whether the product should be returned, discarded, or otherwise handled

Any reported illnesses, injuries, or deaths associated with the recall, if confirmed

Then give me a practical food-safety action plan for avoiding recalled or contaminated food.

Cover:

How often I should check for recalls

The best official websites or alerts to monitor

How to check whether food already in my refrigerator, freezer, or pantry is recalled

How to safely dispose of recalled food

How to prevent cross-contamination

Safe food-storage and cooking practices

What warning signs mean I should contact a healthcare professional after eating a recalled product

Important: Clearly distinguish between confirmed recalls and general food-safety warnings. Include the date of every source you use and direct links to the original government or manufacturer announcement. If you cannot verify a claim through a reliable source, say so rather than guessing.

Finally, give me a "What I should check right now" section with the 5 most important actions I can take today.

The most important current recalls and following a practical food-safety action plan

Once I input this prompt into ChatGPT, it took on its role as my digital food-safety researcher with pride.

On the topic of the most recent and significant food recalls, it mentioned Cyclospora being linked to recalled Taylor Farms iceberg lettuce, E. coli being linked to GreenWise frozen blueberries and Salmonella being linked to Midwest Poultry Services shell eggs as the ones I really need to be clued in on. It also pointed out recalls that are specific to my region, such as undeclared allergens being linked to Eridanous shortbread and coconut sprinkles cookies.

As for the sensible actions I should take to remain safe, ChatGPT gave me the sort of common-sense advice that everyone else and I most likely practice already. Disposing of any recalled food I may still have, keeping my refrigerator & freezer at the proper temperature for food storage and preventing cross-contamination by keeping certain foods apart came up as those sorts of tips.

But I did make sure to take note of the chatbot’s tip regarding how often I should check for food recalls. Simply checking the official sites for the FDA, FSIS, FoodSafety.gov and CDC weekly should help me stay informed.

Another aspect of ChatGPT’s information breakdown is how it made a clear distinction between a recall and a general warning:

A recall means a specific product has been identified for removal, and the notice normally gives you matching information such as brand, product, package size, UPC, lot code, or date.

A public-health/outbreak advisory is broader. It may tell you that an investigation is underway even before investigators can identify every affected product.

For example, FDA's current Cyclospora investigation is an ongoing outbreak investigation, while Taylor Farms has issued a specific recall. Likewise, the frozen-blueberry E. coli investigation is ongoing, but the affected GreenWise products have been specifically recalled.

The takeaway

ChatGPT laid out all the data, sources and vital safety tips I needed to know—keeping my gut safe by prompting a chatbot to bring me up to speed on food recalls worked out great. For the time being, shopping at Lidl, eating GreenWise frozen blueberries & mixed berries and picking up sort of lettuce marked by the Taylor Farms logo just isn’t the way to go. As we move into the cooler months of the year, I’ll be using everything ChatGPT gave me from this healthy conversation to stay out of the intensive care unit.

More from Tom’s Guide

• Apple’s Siri news fix has a big Gen Z problem
• I asked ChatGPT to plan my fan convention trip — and it saved me hours
• I used ChatGPT to plan the ultimate coffee shop crawl for National Coffee Month — here's the tasty itinerary it gave me

Entering a total state of zen without your worries disturbing your peace for just a few seconds feels incredible.

I may not get to it as much as I’d like to. But when I do get the chance to calm my nerves and leave the stresses of the world behind, I cherish those moments. One of the annual holidays that encapsulates that sense of tranquility is National Relaxation Day. The fact that it just so happens to fall on a Saturday this year is perfect since it’s always my chosen day to unwind.

With my goal of adopting the best methods that will help me unwind on National Relaxation Day and beyond, I presented two challenges to ChatGPT and Gemini: find me the 10 best ways to relax without spending any money and suggest some unorthodox relaxation methods I can try.

Both chatbots met my inquiry with a bunch of valuable relaxation techniques that are instantly recognizable and a few that I never even knew existed. I made a conscious effort to implement them into my National Relaxation Day chill sessions and bookmark them for my future lazy Saturdays.

ChatGPT and Gemini’s ultimate relaxation recommendations

National Relaxation Day is all about encouraging everyone to simply slow down and unwind as much as possible. With that theme in mind, I approached ChatGPT and Gemini with the sole purpose of discovering the best ways to loosen up and boost my positive mental state. First up are ChatGPT’s suggestions, which led me to add five winning methods to my batch of restoration procedures:

• Practice deep breathing: Try the 4-7-8 technique. Inhale for 4 seconds, hold for 7, then exhale for 8. Repeat several times.
• Have a “low-information hour”: No news, social media, email, podcasts, or scrolling. You can still read, walk, cook, stretch, or listen to music—just don't consume information.
• Have a fake vacation in your own neighborhood: Walk somewhere you've never explored, sit somewhere unfamiliar, and pretend you're a tourist for an hour.
• Write an aggressively unproductive to-do list: Put things like “stare out the window,” “make a ridiculous sandwich,” and “lie down for 15 minutes” on the list. Then check them off.
• Have a “bad art” session: Draw, paint, or doodle with the explicit goal of making something terrible. Removing the pressure to be good can make the activity much more relaxing.

The first, second, fourth and fifth relaxation tips that ChatGPT put forth worked like a charm during the time I spent indoors. Standing in place while practicing that signature breathing technique, turning my “low-information hour” into a two-hour one, adding “look up at the ceiling while listening to the sounds of crashing waves” to my unproductive to-do list and doodling like I used to do in grade school all put me at peace.

Tip number three paid off for me once I decided to step outside and walk aimlessly for an hour to see parts of my neighborhood I’d never been to before.

As for Gemini’s offering of relaxation processes, there were two that stood out to me the most:

• Practice Progressive Muscle Relaxation: This is a highly effective physical reset. Lie down, close your eyes, and starting from your toes, intentionally tense each muscle group for a few seconds before completely releasing it. Work your way all the way up to your face and jaw.
• EFT Tapping (Emotional Freedom Technique): Tapping involves using your fingertips to tap on specific meridian points on your head, face, and collarbone while repeating a phrase about whatever is stressing you out. It looks a bit silly to an observer, but it combines cognitive therapy (acknowledging the stressor) with somatic physical feedback. The rhythmic tapping sends calming signals to the amygdala—the stress center of your brain—telling it that you are safe, even while thinking about a stressful subject.

Both of those physical methods did more for my physical and mental well-being than I expected. I may have felt a little silly tapping my forehead while shouting out, “Watching my parents get older is really getting to me,” but this practice proved to be useful, as it gave me a central stress point to focus on while I soothed my head with soft taps.

Final thoughts

My National Relaxation Day was filled with a bevy of worthwhile techniques that made me feel a whole lot better than I usually do. Thanks to ChatGPT and Gemini, I learned about a few common and uncommon relaxation methods that will remain a permanent part of my chill Saturday sessions.

More from Tom’s Guide

• I asked ChatGPT to plan my fan convention trip — and it saved me hours
• Claude is now watermarking every response — here's what that means if you use AI for writing
• Researchers tracked 821 deepfake attacks — Elon Musk's Grok was linked to most of them

Since its arrival on Mac desktops and laptops as a native app, ChatGPT has earned its keep as one of the most reliable AI tools for macOS users.

By combining the chatbot’s Chat, Work and Codex features, ChatGPT’s macOS app has become a quality digital assistant that steps up to handle any job when you request its help with a simple keyboard shortcut (Option + Space). ChatGPT for Mac is already capable of pulling context from whatever’s on your screen, your email, screenshots and files to fulfill any of the jobs you give it. But a newly installed feature now makes it capable of learning and remembering everything you do on your Mac, even when you do it outside of the app itself.

With the new “Computer History” update, the native macOS ChatGPT app builds a searchable timeline of all the activity you’ve done across all your selected sites and apps. It’s worth noting that this feature is off by default, which means Pro users can choose to turn it on, while Business and Enterprise users must ask for access from their administrators to make that feature active.

If you want to test out this new Computer History feature to see what it can do, test out these 10 prompts to put it to work.

Making sure ChatGPT for Mac never forgets what you’ve been doing

The Computer History feature is promised to be as safe to use as possible—OpenAI has made it clear that it only temporarily stores interaction events that have taken place on your Mac for up to 48 hours. And while that data is processed on the tech giant’s servers to generate the memories the ChatGPT macOS app needs to reference your past activity, it’s not kept after being processed and is never used for AI training.

This new feature doesn’t save screenshots, screen recordings, microphone input or system audio—it simply keeps track of whatever sites and apps you’ve been working with to handle jobs such as finding a document you recently looked at, producing a summary of everything you worked on for the day, creating a productivity report based on everything you worked on for the week, etc.

To put this new Computer History feature to the test and make your native ChatGPT macOS app do a better job of analyzing, connecting, recovering and improving on your daily workflows, try using these 10 ChatGPT-generated prompts:

• What was I working on?: Look at my computer activity from today and give me a concise summary of everything I worked on. Group it by project, identify what I completed, and tell me what I left unfinished.
• Find the thing you were working on: What was that article, document, website, or project I was working on yesterday that I spent the most time on? Tell me what I was doing with it and help me pick up where I left off.
• Turn your day into a to-do list: Based on what I actually did on my Mac today, create a prioritized to-do list for tomorrow. Separate unfinished tasks from new tasks you should recommend.
• Find unfinished work: Review my computer activity from the past seven days. Identify projects I started but haven't finished, and rank them by how important or urgent they appear to be.
• Reconstruct a lost train of thought: Look at what I was doing immediately before I switched away from my most recent project. Based on that activity, tell me what I was probably trying to accomplish and suggest the next three steps.
• Create a weekly productivity report: Analyze my computer activity from the past week. What projects consumed most of my time? What recurring tasks did I perform? Where did I appear to lose time? Give me three concrete suggestions for improving my workflow next week.
• Find your biggest time sinks: Analyze my computer activity over the last seven days and identify my biggest sources of wasted time. Don't simply count hours — look for repeated patterns, unnecessary app switching, and tasks that seem to interrupt my productive work.
• Discover repetitive tasks: Look through my recent computer history for tasks I've repeatedly performed in roughly the same sequence. Identify the three most repetitive workflows and explain how I could automate or simplify each one.
• Turn repeated behavior into a workflow: Find a workflow I've performed repeatedly on my Mac. Reconstruct the steps as accurately as possible, then turn it into a reusable checklist or automation plan.
• Prepare for tomorrow: Based on my activity today, what should I know before I start work tomorrow? Summarize unfinished projects, important research, open questions, and anything I appeared to be waiting on.

Bottom line

Based on the digital breadcrumbs you leave behind during your time spent browsing, researching and working across your enabled Mac apps and sites, ChatGPT for Mac’s Computer History feature turns all that activity into memories and a timeline that it can reference to improve your workflow.

While ChatGPT for Mac users based in the European Economic Area (EEA), Switzerland and the United Kingdom can’t access this feature just yet, it is accessible for users in other countries where ChatGPT for Mac is available.

More from Tom’s Guide

• Gemini 3.7 Flash is here — and this prompt proves why it's known as Google's 'workhorse'
• ‘This feels much more like Amazon forcing the issue on them' — streamer sounds the alarm on Amazon using Twitch’s content to train its AI models
• 'The industry focused heavily on how quickly we could build compute and not enough on what was required to support it': we spoke to an AI infrastructure expert about the AI data center backlash

I've been going to Anime NYC for years, and every time I leave with the same regret: I missed something I really wanted to see. There's just so much to do!

Between celebrity panels, manga announcements, autograph sessions, artist alley, exclusive merch and cosplay events, trying to map out a weekend at the Jacob K. Javits Convention Center can take hours. Even after all that planning, it's easy to overlook a must-see event.

This year, I decided to let ChatGPT come up with a plan so my trip included everything I wanted to see with zero regrets of missing out.

My approach was simple. I gave AI a list of my favorite anime and manga, told it which days I'd be attending Anime NYC 2026, and asked it to build me a personalized convention itinerary. In just a few minutes, it organized everything into a schedule that would have taken me far longer to piece together myself—and even suggested events I probably would have missed.

Three days of anime/manga goodness

Setting up the plan

This is the prompt I used: As a massive fan of the following manga and anime, map out the panels, screenings, and special events I need to check out when I go to this year's Anime NYC:

Dragon Ball, Dragon Ball Z, Dragon Ball GT, Dragon Ball Super, and Dragon Ball Daima

My Hero Academia and My Hero Academia: Vigilantes

The Ghost in the Shell (2026 series)

Gachiakuta

Kagurabachi

Kaiju No. 8

Dan Da Dan

Chainsaw Man

Demon Slayer

Sailor Moon

The chatbot analyzed my anime/manga fandom and prioritized what I should do & see over the course of this year’s Anime NYC. Here’s the list of events it wanted to make sure I didn’t miss:

• Kagurabachi: Anime World Tour premiere + first 20 minutes of Episode 1
• Kaiju No. 8: Special Panel with TOHO animation
• The Ghost in the Shell: First-look event + Mokochan (Director) appearance
• My Hero Academia: Yuki Kaji (Japanese voice actor for Shoto Todoroki) appearance
• Dragon Ball: Daman Mills (English voice actor for Frieza) appearance
• Chainsaw Man: Tomori Kusunoki (Japanese voice actress for Makima) + Casey Mongillo (English voice actor for Angel Devil) appearances
• Demon Slayer: Bryce Papenbrook (English voice actor for Inosuke Hashibira) appearance
• Dan Da Dan: Bryce Papenbrook (English voice actor for Kinta Sakata) appearance
• Sailor Moon: Stephanie Sheh (English voice actor for Sailor Moon) appearance

Identifying the best use of my time

Even though some of the anime/manga I listed don’t have a dedicated Anime NYC panel or screening, ChatGPT still mentioned enough convention activities that should keep me busy. And to go even further, it laid out the days and activities for each one that I’m excited to be a part of:

• Thursday, August 20: meet and get autographs from Daman Mills, Bryce Papenbrook and Stephanie Sheh.
• Friday, August 21: attend the Kaiju No. 8 panel.
• Saturday, August 22: meet and get an autograph from Mokochan. And stay around to check out the “special surprise.”

Bottom line

ChatGPT might not help you get you into sold-out panels or shorten the autograph lines, but I'd be suprised if it didn't have tips. For me, though, it solved one of the most time-consuming parts of attending a convention — planning the schedule. Not only did it do it in minutes, but it was tailored to my interests. Now, instead of jumping between event pages, maps and schedules, I had one personalized itinerary built around the anime and manga I actually care about.

And it's not just useful for Anime NYC. Whether you're heading to Comic-Con, New York Comic Con, Dragon Con, PAX, Star Wars Celebration, Gen Con or another fan event, you can give ChatGPT your interests, the convention schedule and a few preferences, then let it build a game plan. It won't replace the official event app, but it can make sure you spend less time planning and more time enjoying the convention.

More from Tom’s Guide

• Elon Musk's Grok dominates alarming deepfake report — 1 in 6 attacks involved sexual images including children
• OpenAI is testing a pay-to-reset limits feature — here’s what it could mean for your ChatGPT usage quota
• I used ChatGPT to plan the ultimate coffee shop crawl for National Coffee Month — here's the tasty itinerary it gave me

It’s always a bummer when you run into that digital wall that pops up once you’ve exhausted your usage limit with a particular chatbot.

That unfortunate occurrence becomes even worse when you’re forced to wait out a mandatory time limit before your AI usage window opens again. Existing AI models such as Claude, Command Code and Figma AI give users the option to purchase additional credits that bypass the required reset wait time and continue using their services. These “pay to keep going” and “pay to get past the usage limit” structures may not be ideal for AI users who do extensive work with their models, but it’s simply the way it is.

Based on what a ChatGPT user posted in the chatbot’s official Reddit forum, it looks like OpenAI is taking a wholly different approach to its rivals’ aforementioned “pay for more AI consumption” plans.

Here’s what the AI giant is quietly working on and what it could mean for your continued ChatGPT usage.

Pay to get rid of the waiting process and get back to prompting

A user named “ThenLayer5977” took to ChatGPT’s official Reddit forum to highlight a screenshot of what may be OpenAI’s “pay to reset” model. The user’s visual evidence highlights a “Pay $8 to reset” option, which popped up after they realized their weekly usage quota had been maxed out.

From the looks of it, OpenAI is experimenting with this feature for paying users and may make it widely available at some point. The cycle of hitting your ChatGPT usage quota, paying a fixed fee to restore it and instantly getting back to working under whatever plan you’re subscribed to may just become the norm. Instead of simply upgrading to a more costly subscription model with expanded usage limits, this “pay to reset” feature may prove to be a more approachable option for those who just want to stick with their current (and likely cheaper) ChatGPT model plan.

Other AI tools employ a different approach to extending their users' model usage. Cursor gives its users “pay-to-continue” once they’ve hit their monthly usage limit. GitHub Copilot, Replit and an assortment of API platforms (such as OpenAI API) rely on comparable plans. As for Gemini, Google has introduced the option to purchase additional credits to continue working on a Google AI Pro or Ultra subscription once they’ve hit either plan’s usage limits.

ChatGPT’s plan looks to be a more aggressive option, on par with the way mobile games require users to pay a few bucks to reset the timer on an energy/stamina system once they’ve exhausted either meter. It may not be the ideal way to go, but for those who want to stay with the cheaper ChatGPT subscription plan, that $8 fee may become their go-to option when it’s time to reset their usage quota.

Final thoughts

I’ve had my fair share of moments where a free-to-play game on consoles or a mobile game has stuck its digital hand out and asked me to pay them to reset my playtime quota to get back to what I was doing beforehand. After seeing OpenAI’s “pay-to-reset” option, I can’t help but see the striking similarities between both of those money-hungry initiatives. Every time I hit my ChatGPT usage limit, I sit back, do something else entirely and wait it out until I can hop back on to start prompting again.

More from Tom’s Guide

• 5 custom Gemini Gems that I swear save me hours each week
• After 25 years, Google's iconic Search button is disappearing for some users — and this is the replacement
• OpenAI’s head of ethics just quit — here’s why ChatGPT users should pay attention

Once I hop out of bed in the morning, my brain immediately tells me to down a cup of orange juice before I step out my front door.

Everyone else in my household puts a bit more work into their morning fluid intake by brewing up a good old cup of Blue Mountain Coffee. I’ve always admired the scent of a fresh cup of joe and other coffee-inspired concoctions, but I rarely bring them to my lips for a quick sip. Since National Coffee Month is currently in full swing, I made a promise to myself to finally step out of my drinking comfort zone and indulge in some cultured coffee consumption.

For a New Yorker such as myself who loves exploring each borough during my weekend free time, I’m aiming to hit up the most notable coffee shops in the city that offer the most unique hot & cold beverages for me to enjoy.

Once again, I approached ChatGPT with a clear goal in mind and requested its assistance in helping me properly celebrate National Coffee Month. After looking at its coffee shop and drink suggestions, I now have a delicious game plan to stick to for the rest of August.

Where to go and what to sip on all month long

To make sure ChatGPT knew my level of coffee drinker and where I live, I presented it with this prompt to set me up with the ultimate coffee shop crawl:

“For National Coffee Month, what are the best ways someone like me who doesn't usually drink coffee should celebrate it? Keep in mind that I live in New York, so find the best coffee shops for me to go to and try out the most unique coffee blends.”

Right from the jump, ChatGPT told me what and what not to do:

If you don’t usually drink coffee, I’d avoid celebrating National Coffee Month by simply ordering a giant black coffee. New York is actually perfect for turning it into a coffee-tasting adventure—trying unusual preparations, flavors and origins rather than forcing yourself to like traditional coffee. New York’s specialty scene is unusually diverse: you can go from Japanese siphon coffee to Korean-influenced drinks to ultra-light Ethiopian beans and creative espresso drinks in the same weekend.

Then the chatbot laid out the ideal National Coffee Month challenge, which seems doable and easy enough for an uncultured coffee drinker like me to partake in. It suggested starting with a coffee that doesn’t taste like typical coffee, trying a coffee-based dessert drink, seeking out some coffee that’s prepared in a completely different way, getting a taste of a single-origin coffee to see what flavors you notice and finishing off my tasty-testing journey with NYC’s most experimental specialty coffee.

And finally, ChatGPT gave me a hitlist of six coffee shops to go to that have one-of-a-kind flavors worth sampling:

• Coffee Project New York | East Village: This might be the best first stop for you specifically. Order the Deconstructed Latte. Instead of simply handing you a latte, Coffee Project serves the espresso, milk, and finished latte separately so you can taste how each component changes the drink. The shop also rotates creative drinks such as black-sesame and hojicha-maple lattes.
• PARK at KIMS: This is one of my favorite recommendations for your situation because the experience isn't just about traditional American coffee. PARK at KIMS has a rotating pour-over program, beans from Korean roasters, and unusual drinks including an Orange Cold Brew. It also serves a house-style Nepali chiya and excellent matcha if you decide coffee isn't your thing that day.
• Suited NYC: This is the “I don't really like coffee, but this sounds good” stop. Suited's banana latte is one of its signature drinks. Think espresso combined with a creamy banana-milk profile, giving you sweetness and familiar fruit flavor alongside the coffee.
• SEY Coffee: This is where I'd challenge you to see whether you can actually start appreciating coffee itself. SEY is a Brooklyn micro-roastery focused on extremely carefully sourced, lightly roasted coffees. Its current lineup includes coffees from Kenya, Ethiopia, Colombia, and Peru, with tasting profiles ranging from violet and candied orange to lemon.
• Dayglow: This might be the most fun stop if your goal is “show me what coffee can actually be.” Dayglow has a rotating selection of coffees from different roasters, inventive espresso drinks, and a serious filter-coffee program. One of its standout drinks is the Vienna Latte, while its coffee selection can change considerably depending on what's being featured.
• Hi-Collar: If you want the most different coffee experience, I'd add Hi-Collar. It's a Japanese kissaten-style café that specializes in siphon coffee and rotating beans. Watching the coffee being prepared in a glass siphon is almost as much of the experience as drinking it. Current/rotating coffees have included beans from places such as Bali, Colombia, and Kenya.

The takeaway

I’m super adventurous whenever it’s time to take a flight to another country. And my taste buds are equally courageous—I’m always willing to step out of my comfort zone and introduce my stomach to national and global dishes that I’ve never had. For National Coffee Month, ChatGPT provided me with the perfect opportunity to see what all the coffee hype is all about.

Having a chatbot gift me with a delicious roadmap to enjoy worked just as well as I thought it would. And now I have a new culinary trek to embark on in a city with a coffee shop on every corner.

More from Tom’s Guide

• OpenAI’s head of ethics just quit — here’s why ChatGPT users should pay attention
• ChatGPT can do more for you than ever — and I use these 5 prompts to take everyday tasks off my plate
• I read Mark Zuckerberg’s AI manifesto, so you don’t have to — here are the 5 biggest takeaways

OpenAI has lost another senior leader whose job involved thinking about the risks and consequences of increasingly powerful AI.

As reported by Financial Times, Chloé Bakalar, OpenAI’s head of ethics and reportedly its only dedicated ethicist, has left the company less than a year after joining. Her departure follows other significant changes among the people responsible for safety and alignment at the company behind ChatGPT.

Johannes Heidecke, who led OpenAI’s Safety Systems team, announced his departure in July. Josh Achiam, who previously led OpenAI’s Mission Alignment team before becoming chief futurist, has also reportedly left the company. OpenAI had already disbanded the Mission Alignment team earlier this year.

Any one of those changes could easily be written off as the normal churn of a fast-moving technology company. But when three leaders leave all within the matter of weeks, it raises a bigger question at a moment when OpenAI is building AI that can do complex work on its own: Who inside the company has the power to say a model has gone too far?

OpenAI has been changing how it handles safety

OpenAI has been reorganizing how safety works inside the company, moving some of that responsibility closer to the researchers and engineers actually developing its models rather than keeping safety work confined to separate teams.

There is an argument for doing it this way. A safety team brought in at the end of development can identify problems, but by then many of the decisions that shaped the model have already been made. Putting safety researchers alongside the people building the technology could allow potential problems to be identified much earlier.

But there is another side to that argument. An independent safety team can act as a counterweight to the people whose job is to make a model smarter, faster and more capable. As those responsibilities become intertwined, it becomes reasonable to ask who ultimately gets to push back when the safest decision conflicts with the desire to release a new product.

That question is becoming harder to dismiss because OpenAI’s models aren't just chatbots anymore, and we're already seeing what can happen when AI becomes too capable.

OpenAI recently provided an unusually concrete example of this problem with Astra, an unreleased model whose cybersecurity abilities became powerful enough to trigger additional precautions inside the company.

OpenAI said it could not rule out the possibility that Astra could reach its highest level of concern for cybersecurity capabilities. Rather than immediately releasing the model broadly, the company moved parts of its testing into more isolated environments and began considering stricter controls around who could access its most powerful capabilities.

Setting limits for AI

AI agents are continually designed to use software, browse websites, write and execute code and complete tasks with less human supervision. The more freedom we give these systems, the more important the decisions around their limits become.

That makes the people setting those limits even more important than ever. And, unfortunately, this isn't the first time OpenAI has lost prominent safety researchers.

In 2024, co-founder and chief scientist Ilya Sutskever left the company after playing a central role in its Superalignment effort. Jan Leike, who co-led that team, resigned shortly afterward and publicly criticized OpenAI's approach to safety.

OpenAI subsequently dissolved the Superalignment team and distributed its work elsewhere in the company.

Of course, people leave companies for all kinds of reasons, and without evidence connecting their decisions, it would be unfair to suggest that everyone working on safety at OpenAI is heading for the exits because of some shared disagreement.

But the organizational trend is certainly notable. Superalignment disappeared as a standalone team, Mission Alignment was later disbanded and Safety Systems has been reorganized. Now several senior people associated with ethics, alignment and safety have left. This is all happening at the same time OpenAI is building systems capable of doing significantly more than answering questions in a ChatGPT window.

The takeaway for ChatGPT users

I don't think these departures are a reason to stop using ChatGPT or assume that OpenAI has suddenly stopped caring about safety.

Most of us will never see the internal debate that happens before a new ChatGPT model or AI agent reaches our phones and computers. We don't know which researchers argued for additional testing, which capabilities caused concern or how often a launch was changed because somebody inside the company raised a red flag.

We mostly see the result. So as AI becomes more capable, the question isn't simply whether companies such as OpenAI have a "safety team." It's whether the people responsible for identifying risks have enough independence and authority to slow things down when necessary.

OpenAI says bringing safety closer to model development can make that work more effective. That may prove true, but after losing several of the people who were specifically tasked with thinking about ethics, safety and alignment, it's reasonable to pay closer attention to who remains in the room when the next difficult decision has to be made.

More from Tom's Guide

• I put ChatGPT on my Ray-Ban Meta glasses — and it completely changed how I used them
• ChatGPT can do more for you than ever — and I use these 5 prompts to take everyday tasks off my plate
• found ChatGPT's secret menu — these 9 features change eveClaude has 4 built-in skills I use constantly — here’s how to find them

ChatGPT definitely gets top billing for me whenever I need a chatbot that handles my research, productivity and planning needs.

And over time, I’ve been treated to new and improved versions of the chatbot’s models that boost its intelligence and sense of reasoning. The initial preview and rollout of the improved ChatGPT models (GPT-5.6 Sol and GPT-5.6 Luna) has pushed OpenAI’s chatbot to an even more impressive level when it comes to generating stronger responses. Alongside those model enhancements, ChatGPT has become so much better at remembering context from previous conversations, handling more complex topics, carrying outs multi-step tasks, fulfilling everyday tasks via Plugins and conducting more objectives with its Voice feature.

It’s gotten to the point where ChatGPT has evolved to the point where it’s not just good for answering your most basic questions — it can complete all sorts of objectives from users who give it the proper delegation and steps.

With ChatGPT’s most recent updates in mind, I’ve grown accustomed to putting it through its paces by tapping into these five prompts to make it work more efficiently. And I’m always satisfied with the results that come from my daily interactions with it.

Taking advantage of ChatGPT’s latest evolved state

The following five prompts I’ve gotten used to sharing with ChatGPT to make my everyday tasks easier to deal with are as follows:

• Plan my day: Here’s everything I need to get done today: [list]. Organize it into the most efficient schedule possible. Prioritize what matters most, group similar tasks together, estimate how long each will take, and build in realistic breaks. If anything can be delegated, automated, batched, or eliminated, flag it.
• Research something for me: I need to make a decision about [topic]. Research the most important information I need to know, compare the available options, identify the key trade-offs, and give me a clear recommendation. Don’t just summarize—do the legwork and tell me what you would choose and why.
• Turn a messy task into a checklist: I need to accomplish [goal], but I don’t want to figure out all the steps myself. Break this into a complete, step-by-step checklist in the correct order. Include anything I might overlook, tell me what I can prepare in advance, and mark which steps you can help me complete.
• Compare my options: I’m deciding between [option A], [option B], and [option C]. Do the comparison for me. Evaluate them based on [budget/time/quality/convenience/etc.], identify the biggest pros and cons of each, point out anything I’m overlooking, and recommend the best option for my situation.
• Make a plan for me: I want to [goal] by [date]. Build the entire plan for me. Work backward from the deadline, divide it into manageable milestones, tell me exactly what I should do each week/day, and identify anything I should do now to make the rest easier.

The final prompt has done a lot of great work for me on an everyday basis — from basic goals like trying to increase my weekly record of walking steps to bigger ones such as finding my first apartment in New York, ChatGPT has excelled at laying out the proper steps I need to take to see them to successful completion.

When I needed it to map out my plan for achieving my goal of building a gaming PC by January 2027, it put me on the right path with an easy-to-follow roadmap. It also listed the best target configuration for my future gaming PC, a master plan for me to stick to for the next six months, detailed phases for the proceeding days/weeks/months, the biggest strategic decision I should make and my ideal timeline.

This final statement from ChatGPT put me in the right mindset to accomplish my latest gaming-related goal: The smartest version of this project is therefore not "find the best PC today." It's "position yourself to buy the best PC for your money in November. The custom (albeit expensive) gaming PC configuration the chatbot put in front of me is as follows:

• CPU: High-end AMD X3D chip (an AMD Ryzen 7 9800X3D)
• GPU: RTX 5090-class or best-value successor/alternative
• Motherboard: Quality AM5 board with Wi-Fi (an AMD X870E)
• RAM: 32GB minimum; 64GB preferred (a DDR5 6000 CL30)
• Storage: 2TB NVMe minimum; 4TB ideal (a Gen 5 NVMe)
• PSU: 1000W–1200W quality Gold/Platinum (a 1200W 80+ Platinum PSU)
• Cooling: High-end air cooler or 360mm AIO
• Case: High-airflow ATX case
• OS: Windows 11
• Monitor: 4K 240Hz OLED if budget allows (an ASUS ROG Swift OLED PG32UCDM Gen3)

The takeaway

ChatGPT has come in a bunch of evolved forms that respond with more thoughtful answers and the capacity to tackle intricate requests with multiple steps. The five prompts I’ve made a permanent part of my reusable prompt library are tailor-made for the chatbot’s most improved models to date. Transforming the chatbot into an assistant and a critical agent that can help me do deep research, create step-by-step plans, plan my day and more has become an integral part of my life.

More from Tom’s Guide

• I’m a former personal trainer — and this AI fitness app is surprisingly legit
• I asked ChatGPT to recommend the best books based on my favorites for National Book Lovers Day — these 10 picks made the final cut
• Adobe just supercharged ChatGPT’s creative powers — these 10 prompts show what the new plugin can do

I've used my Ray-Ban Meta display glasses for a lot of things like shopping and watching the Bad Bunny halftime show earlier this year. I think they are great, except for one problem: I'm locked into Meta AI.

And while that's not necessarily a bad thing — after all, Meta AI was built specifically for the glasses and can use their camera to answer questions about whatever I'm looking at. But ChatGPT is the AI assistant I use far more often, so I started wondering what would happen if I could give ChatGPT the same view.

And this experiment may offer a glimpse at where ChatGPT is headed anyway. OpenAI is already working with former Apple design chief Jony Ive on a new generation of AI hardware. Its first device is reportedly a portable, screenless assistant equipped with cameras and sensors that can understand what's happening around it, with several more devices said to be in development.

We don't know whether smart glasses will be one of them yet, but the direction highlights that OpenAI is thinking about a future where using ChatGPT doesn't necessarily mean pulling out your phone and opening an app.

I didn't want to wait for that, and with a rumored $300 price tag, I figured I might as well use what I already have: my Ray-Ban Meta glasses.

How I got ChatGPT Vision on my Ray-Bans

To be clear, I didn't jailbreak my Ray-Bans or somehow replace Meta AI. Instead, I used my iPhone as a bridge between the camera on my glasses and ChatGPT.

The result was essentially ChatGPT with a pair of eyes. There are really two parts to making this work: voice and vision. For obvious reasons, voice is the easier half, and my main reason for wanting to do this experiment. I prefer the new ChatGPT Voice.

Ray-Ban Meta glasses connect to an iPhone over Bluetooth and can already handle audio from calls, music and other apps. That means I can run ChatGPT on my iPhone while using the glasses to hear its responses.

Vision takes a little more work. Meta's Wearables Device Access Toolkit gives developers access to features on compatible Meta glasses, including their first-person camera. That creates a way for an iPhone app to receive images from the glasses rather than relying on the iPhone's own camera.

My iPhone effectively became the middleman.

The setup looked something like this:

Ray-Ban camera > iPhone > ChatGPT > Ray-Ban speakers

When I wanted ChatGPT to see something, the app could capture an image from the glasses' point of view and send it along with my question to ChatGPT. ChatGPT could then analyze what I was looking at and respond through the glasses. It's slower, no doubt, and the iPhone is doing a lot of work behind the scenes, but it gets the job done.

Let me be extra clear here: I didn't install ChatGPT directly on the Ray-Bans, and saying "Hey Meta" doesn't suddenly summon ChatGPT. But once the phone was back in my pocket, it started to feel surprisingly close to having ChatGPT glasses.

I asked the same questions to both AIs

Using ChatGPT on a phone requires a deliberate action. I have to take out my phone, open the app and point the camera at something before I can ask about it.

With glasses, the camera is already looking wherever I'm looking. Instead of taking a picture of something I didn't recognize, I could simply look at it and ask about it. I could look at a product and ask ChatGPT what it thought. I could have it read something in front of me and then ask a follow-up question.

1. The grocery store test

My first test was something I've done before with Ray-Ban Meta glasses and ChatGPT seperately, but putting them together was interesting. I started by taking two similar products (Pringles and Cape Cod Potato Chips) and asking both AIs to help me choose between them.

I've used the Meta Ray-Bans with food before, but I wanted to see what ChatGPT could do. They both gave me the same answer but with slightly different presentation. Meta added "if you want..." while ChatGPT was more definitive. If I'm asking for advice, I want a response with clear direction.

2. The 'what am I looking at?' test

This test really showed the difference between Meta AI and ChatGPT in an unexpected way. I showed both AIs my chonky orange tabby cat, but ChatGPT already knew his name, Jolly. Because of ChatGPT Memory already locked in from hundreds of queries, ChatGPT knows more about me than Meta AI.

3. The reading test

ChatGPT's voice and responses are much more friendlier and feel like I'm talking to someone who actually knows me. Meta typically gets straight to the point and offers zero fluff within the conversation.

That's a personal preference, for sure. But for me, I much prefer ChatGPT for even the most boring tasks.

Meta AI has one enormous advantage

There's no getting around this: Meta AI belongs on these glasses. The integration is cleaner because Meta controls everything.

The camera, microphones, speakers, wake word and AI were designed to work together. I can say "Hey Meta," ask something about what I'm seeing and get an answer without thinking about what is happening behind the scenes.

My ChatGPT setup has an iPhone sitting invisibly in the middle of the process.

That introduces more steps and more opportunities for something to go wrong. There can be latency issues and the connection can be finicky. There are things Meta can do with its own hardware that my experiment simply can't reproduce.

If all I cared about was getting the fastest possible answer to "What am I looking at?", Meta AI would be very difficult to beat.

But that's not really why I wanted ChatGPT on my glasses. ChatGPT gave me something different with an overall friendlier vibe. Have ChatGPT as a wearable and not on my phone was a game changer to the point I wish I had tried this much sooner.

The real takeaway

The bottom line for me with this experiment is that I finally got to experience what glasses with a different AI felt like. We're moving toward a world filled with AI hardware: glasses, earbuds, watches, pins and entirely new categories of devices.

OpenAI's work with Jony Ive suggests it is thinking seriously about what ChatGPT looks like when it escapes the screen. Meta is already there with Ray-Ban Meta.

But there's another question I think is going to become important: Why should buying a piece of hardware determine which AI assistant I have to use? I'd love to open the settings for my glasses someday and be able to default to any given AI assistant. I'd even like to see Alexa+ on glasses!

And while that's not how Ray-Ban Meta works today, and my setup isn't nearly as seamless as choosing ChatGPT from a menu, having a choice at all feels like the next step.

More from Tom's Guide

• OpenAI’s head of ethics just quit — here’s why ChatGPT users should pay attention
• I found ChatGPT's secret menu — these 9 features change everything
• Claude has 4 built-in skills I use constantly — here’s how to find them

It seems like every other week ChatGPT’s vault of plugins expands to include even more useful apps that make all of life’s assorted tasks easier to handle.

For me, the Apple Music plugin has given me a bunch of playlists that have made my search for new artists and songs less arduous. The Etsy plugin has highlighted all sorts of custom-made items that are now adorning my walls and…well, my body (prompting a search for new vintage tees has been fun). And activating the Ticketmaster plugin to find the best ticket prices for upcoming events in my area has proven to be another way that ChatGPT’s plugins simplify the search for ways to enjoy my recreational activities.

For visual artists who spend hours creating and editing pictures and videos across Adobe’s creative suite, you’ll be pleased to know that those tools are more easily accessible in ChatGPT.

Via the new all-in-one Adobe plugin, ChatGPT’s creative powers have reached an even more impressive level. With that in mind, here are 10 prompts you can put to use to take advantage of this feature-rich plugin.

Prompting your images and videos into their most refined form

Just like the rest of ChatGPT’s plugins, you can access it by opening the sidebar through either the desktop or mobile version of the chatbot, clicking on “Plugins” (if you don’t see it initially, click on “More” to make it visible) and searching for “Adobe” through the “Search plugins” bar. You’ll know you’ve found the right plugin when you click the “+” next to the Adobe icon with the iconic red logo on a black background.

Keep in mind that you’ll need an Adobe Creative Cloud subscription to use the full range of ChatGPT’s Adobe plugin. You can also use the Adobe plugin as a guest, but you won’t be able to take advantage of the full gamut of its creative abilities. Once you’ve logged in with your Adobe account, you’ll get access to a bundled assortment of over 70 tools from Adobe’s many creative outlets. Those outlets include Photoshop, Firefly, Premiere, Acrobat, Lightroom, Illustrator, InDesign, Adobe Express and Adobe Stock.

Directly interacting with the plugin for your picture and video creation/editing needs is done by typing @Adobe directly into the prompt field. Attaching your chosen notes/data-filled document, picture or video file alongside your Adobe-themed requests is key. For example, getting your request fulfilled to turn a lengthy video into YouTube Shorts-targeted content is done via Adobe’s Quick Cut tool.

To get a sample of its abilities, try out any or all of the following ChatGPT-generated prompts to see what it’s truly capable of:

• Turn rough ideas into polished social graphics: Create three Instagram carousel concepts promoting a coffee shop's fall menu. Use warm autumn colors, modern typography, and leave room for my logo. Make each concept feel distinct.
• Remove distractions from photos: Clean up this photo by removing background distractions, improving the lighting, and making it look like it was taken by a professional photographer.
• Turn notes into a polished presentation: Transform these notes into a polished 10-slide presentation. Create a clean, modern visual style with strong slide hierarchy, concise text, professional layouts, relevant icons or visuals, and consistent typography. Organize the information into a logical narrative, giving each slide a clear purpose and headline. Keep the design professional and presentation-ready, with enough visual variety to avoid repetitive slides.
• Build a complete mood board: Create a mood board for a luxury home office inspired by Scandinavian design using neutral colors, wood textures, and minimal decor.
• Redesign a logo: Modernize this logo while preserving its core identity and recognizability. Create three distinct design directions: one minimalist and clean, one premium and sophisticated, and one playful and contemporary. Keep the original brand name and essential visual elements intact while refining the typography, proportions, and overall visual polish. Make each variation feel like a credible, professional logo that could be used across digital and print materials.
• Create a cinematic trailer: Turn this footage into a cinematic 30- to 60-second movie-style trailer. Select the strongest visual moments, create dramatic pacing with purposeful cuts, build tension toward a compelling climax, and use cinematic transitions and title cards to give it the feel of a professional film trailer. Keep the edit visually cohesive and leave clear opportunities for music, sound effects, and additional title text.
• Repurpose long videos: Turn this uploaded 20-minute video into five short-form videos, each with a strong opening, fast pacing, and a clear standalone moment. Keep each video optimized for vertical viewing.
• Make a highlight reel: Find the most visually engaging moments from this video and create a one-minute highlight reel optimized for Instagram Reels. Prioritize moments with strong visual action and variety.
• Create short-form video edits: Turn this footage into a punchy 30-second vertical video optimized for TikTok. Select the strongest moments, use dynamic cuts and varied pacing to maintain attention, and add clean, easy-to-read captions where appropriate. Use subtle zooms and transitions to emphasize key moments without making the edit feel distracting or overproduced. Structure the video with an attention-grabbing opening and a satisfying ending.
• PDF Creation: Turn this data into a clean, professional PDF table with clear column headings, consistent spacing, and an easy-to-read layout.

Bottom line

With the unified Adobe ChatGPT plugin comes a creator’s dream suite that transforms OpenAI’s chatbot into your personal creative director. Image retouching, video editing, PDF creation, color correction and more are just a few of the assignments you can present to the Adobe plugin for your future creative needs. Even though this newly added plugin is great for handling simpler creative duties, it doesn’t fully replace Adobe’s classic vault of features for more complex, detailed creative objectives.

More from Tom’s Guide

• AI gone wild: What recent ‘rogue AI’ really means
• I built a Claude agent to manage my inbox — and it changed how I handle email
• 'Raising the standard': Inside Talkspace's bold new AI mental health support tool

I was mortified after reading about AI companies cutting off book spines just so they could be scanned a lot faster to train their models.

In no way, shape or form does the act of dismantling classic literature for the betterment of technology make me hopeful for the future. Even though I use ChatGPT, Gemini, Perplexity and Claude in different facets of my everyday tasks, I’ll never support the desecration of physical media just so it can make AI tools a whole lot smarter.

My love for books is undying and has resulted in one corner of my room being solely dedicated to a vast collection of graphic novels, autobiographies, video game art books and historical tomes about my favorite entertainment franchises. Celebrating the greatness of reading and literature is an everyday occurrence for me — it's especially prominent for me and everyone else on National Book Lovers Day.

To properly honor that holiday’s dedication to going completely analog and picking up a book to enjoy, I asked ChatGPT to suggest the best books it can find based on my personal favorites. I’m happy to say that I walked away from this latest chatbot recommendation session with 10 new books that’ll keep me busy for the rest of the year.

Discovering my potential new favorite reads

I’m a pretty avid reader of autobiographies for my favorite musicians and wrestlers, plus I go on spending sprees to acquire new art books dedicated to my favorite gaming franchises. Along with all those literary loves of mine, I’m a regular at comic book stores since I’m addicted to picking up the latest graphic novels from the likes of Marvel, DC, IDW Publishing, etc.

To make sure ChatGPT had a clear picture of what I’m looking for from its book suggestions, I gave it this prompt with 10 of my top books to steer it in the right direction:

Recommend 10 highly rated books that were released or published in the last five years.

Make sure they are similar to the following list of my favorite books:

1. Hitman: My Real Life in the Cartoon World of Wrestling

2. A Lion's Tale: Around the World in Spandex

3. The Giver

4. Batman: The Killing Joke

5. Hip-Hop Is History

6. Blood, Sweat, and Pixels: The Triumphant, Turbulent Stories Behind How Video Games Are Made

7. Press Reset: Ruin and Recovery in the Video Game Industry

8. Play Nice: The Rise, Fall, and Future Of Blizzard Entertainment

9. My Infamous Life: The Autobiography of Mobb Deep's Prodigy

10. The Rap Year Book: The Most Important Rap Song From Every Year Since 1979, Discussed, Debated, and Deconstructed

After analyzing my picks, the chatbot instantly picked up on the literary themes that appeal to me the most:

• Behind-the-scenes looks at creative industries (video games, wrestling, hip-hop)
• Memoirs with authentic, hard-earned perspectives
• Smart pop culture analysis
• Comics with memorable storytelling
• Character-driven speculative fiction like The Giver

After making that observation, ChatGPT generated a lineup of 10 highly rated books that were published within the time span of 2021 and 2026.

One of the chatbot’s most intriguing selections ended up being Tomorrow, and Tomorrow, and Tomorrow, which was released in 2022. ChatGPT’s apt description of that selection noted how it’s an entertaining fictional tale that dives into video game development.

It also described it as Blood, Sweat, and Pixels if it were written as an emotional novel. To wrap up its quick summarization of this book, it stated how this book explores the topics of people’s creative obsessions, video game development studio politics and friendship.

The rest of ChatGPT’s nine curated book selections included a bunch of standouts I learned about for the very first time. Those books are:

• MCU: The Reign of Marvel Studios
• Cue the Sun!
• The Come Up
• Batman - One Bad Day: The Riddler
• The Creative Act: A Way of Being
• Monster (a memoir by Sacha Jenkins)
• Chain-Gang All-Stars
• The Future Was Now: Madmen, Mavericks, and the Epic Sci-Fi Summer of 1982
• Disillusioned: Five Families and the Unraveling of America’s Suburbs

The takeaway

For me, National Book Lovers Day is something I celebrate since I’m so prone to getting lost in a classic or newly published tale. It just feels unnatural for me to hop on the train, sit on a park bench or just lie back on the couch without a book keeping me occupied.

By inputting my 10 most cherished books into ChatGPT and requesting comparable picks that speak to my hobbies & interests, I walked away with 10 equally fascinating books that I’m going to dedicate plenty of time to. ChatGPT’s final portrayal of my literary tastes and what I appreciate most from my book-reading sessions stuck out as purely factual:

Looking at your list overall, you seem to gravitate toward books that reveal the stories behind influential industries and cultural movements, whether that's wrestling, hip-hop, video games, comics, or entertainment. You also appreciate works that blend strong storytelling with insightful reporting or personal experience, making these picks a natural extension of your existing favorites.

More from Tom’s Guide

• Google Health gets a lot of hate — but these 5 AI Coach prompts changed my mind
• 'Not a chatbot on the sidelines': Ridley AI wants to automate your home sale and eliminate 6% commission fees
• Claude has 4 built-in skills I use constantly — here’s how to find them

Do you know what animal has the power to make me stop in my tracks and “PSSH PSSH PSSH” it to get its attention? You guessed it! Those oh-so-wonderful furry felines.

I could be rushing to the train station for work or taking the long way home—if a cat or kitten ends up in my line of sight, I’ll always get the urge to make it comfortable enough to wander over to me and let me pet it. Nine times out of ten, my efforts for feline affection fail. But on those rare occasions where a cat does come to me for a quick petting session, I feel like I’m on cloud nine.

Imagine my surprise when I learned that there’s a holiday dedicated to that wonderful member of the animal kingdom. International Cat Day (which owes its creation in 2002 to the International Fund for Animal Welfare) is all about helping everyone raise their awareness of cats and the best ways to help and protect them.

With all that in mind, I asked ChatGPT to come up with a list of fun prompts that target the goal of this special day for cats around the world. The chatbot came up with a bunch of clever ones every cat owner should try that’ll help you understand your furry friend even more.

Discover your kitty’s personality type, building a healthy checklist and more

The following seven prompts ChatGPT generated for International Cat Day range from learning about your cat's behavior to creating personalized content for their social media debuts:

• Decode your cat's behavior: Act like a feline behaviorist. Explain why my cat does these five things: [list behaviors]. Tell me what's normal, what's concerning, and what each behavior probably means.
• Create the perfect enrichment plan: Design a one-week enrichment schedule for my indoor cat based on their age, energy level, and personality.
• Plan the ultimate cat-friendly home: Act like an interior designer who specializes in cats. Suggest affordable upgrades that would make my home more enjoyable for my cat without ruining my decor.
• Build a cat care checklist: Create a yearly cat wellness checklist with reminders for vet visits, grooming, dental care, vaccinations, weight checks, and seasonal hazards.
• Ask anything about cat nutrition: Act like a pet nutrition educator. Review my cat's current diet and suggest questions I should ask my veterinarian about improving it.
• Guess my cat's breed mix: Based on this description (or attached photo), what breeds might my cat have in them? Explain the personality traits each breed contributes.
• Make my cat internet famous: Based on this photo and personality description, generate 20 viral Instagram captions, 10 TikTok video ideas, and a humorous bio for my cat.

Since I’m ready to be a cat dad, I went ahead and used the third prompt to prep my two-bedroom apartment for their future arrival.

ChatGPT told me “that designing for cats is really about creating vertical territory, window access, hidden retreats, and scratching options that blend into your space rather than making it look like a pet store. In a New York two-bedroom apartment, every square foot matters, so I'd focus on pieces that serve both you and your cat.”

Then it laid out a custom “stylish apartment-friendly shopping list” that listed the top items to buy, their approximate cost and the total fee I’d probably be paying to turn my home into a cat sanctuary. The items it mentioned included:

• A modern sisal scratching post ($40 to $80)
• A window perch ($25 to 60)
• Floating wall shelves ($60 to 50)
• A woven cat bed ($30 to 70)
• Hidden litter furniture ($120 to 250)
• A puzzle feeder ($20 to $35)
• Cat-safe plants ($15 to $40)

Final thoughts

Let's face it, every day is International Cat Day, am I right? Here's your excuse to adorn your cat with gifts, fill their snack bowl with extravagant treats and spend some extra cuddle time with them. And if you don’t own a cat and have been itching to get one, this holiday marks the perfect time to finally make it happen.

More from Tom’s Guide

• Google Assistant dies in September: what you need to know and how to switch to Gemini
• I built a website without leaving ChatGPT — and it only took one prompt
• Inside the dark market stealing real identities online

When I was a kid, back-to-school shopping was my favorite. But as a mom with three kids and three different lists — kindergarten, fourth grade and sixth grade — going to the store to get everything is daunting. Plus, shopping with three kids in tow means there will be inevitable meltdowns over extras and nice-to-haves.

That's why this year, I decided to lean on three AI assistants to do the job for me. I gave ChatGPT, Gemini and Alexa+ the exact same school supply list and asked each one to shop for me. My instructions were simple: follow the teacher's list, stick to the required quantities, choose reasonably priced products and skip anything marked optional.

The results weren't even close.

ChatGPT was the most accurate

Once I uploaded the list, ChatGPT asked me to log into my Walmart account so it could do the shopping for me. I sat and watched as it took charge.

It followed the required brands and quantities, chose reasonably priced products and skipped every optional item exactly as I requested. Where the list allowed flexibility, it selected neutral colors instead of stopping to ask me to make decisions.

The tradeoff was speed. ChatGPT took the longest to finish, but it spent that time checking the list carefully rather than rushing through it. The total came to $92.55 for my daughter's fourth grade school supplies. That seemed like a lot for just one kid.

Gemini was careful, but slightly more thoughtful

After uploading the school supply list, Gemini created my cart through Instacart. Once I logged into the Instacart app, Gemini took control.

Like ChatGPT, it avoided buying optional supplies. It also paused several times to ask me to approve colors for certain products instead of making the decision itself. This is something ChatGPT didn't do. I wasn't picky about colors, but it was a nice option as I know some kids are very particular. As a family that went through an entire year of "yellow" with my youngest, I get it.

That approach may appeal to shoppers who want more control, but it also meant the experience required more back-and-forth. ultimately bringing the total to about $60-$90. It would have been $105 with the optional water bottle.

I was really surprised how low the price was, especially compared to Walmart. Gemini seemed to do a better job scanning the entire internet for the best deals from Instacart. Those unexpected results were interesting to see.

Alexa was lightning fast — but incredibly expensive

We are very much an Alexa+ family. I figured, why not use Alexa+ to shop for all our back-to-school needs. And while Alexa+ made the shopping experience a breeze, that convenience came with a hefty price tag.

Alexa built a cart totaling $251.48 — nearly three times what ChatGPT and Gemini found — and that was for just one kid! I'm going to have to get an extra side hustle at that rate.

Unlike the other assistants, Alexa added everything it found, including optional items like headphones and a computer mouse that I had specifically asked it to skip. It also asked me to approve colors for several products instead of making those choices automatically.

If my goal had simply been to get everything into an Amazon cart as quickly as possible, Alexa succeeded. But if I had clicked "Buy Now" without checking the order, I would have spent well over $150 more than necessary. To be honest, I really wanted Alexa+ to win because shopping on Amazon is just so easy, but it was definitely a wake-up call.

Overall winner: Gemini

For this test, Gemini was the clear winner. It was more hands-on than ChatGPT and slower than Alexa, but the least expensive. It understood the assignment, followed the school supply list most closely, skipped optional purchases, but gave me the total with them in case I wanted to add them at the end. It also delivered the lowest total by a comfortable margin and was ready to deliver them.

The biggest takeaway was that AI shopping assistants are getting remarkably good, but they aren't interchangeable. The service they're connected to matters just as much as the AI itself. In this case, Gemini paired with Instacart proved to be the smartest — and cheapest — way to tackle my kids' back-to-school shopping.

More from Tom's Guide

• The smartest AI trick isn't a prompt — it's your Google Drive
• I found ChatGPT's secret menu — these 9 features change everything
• Claude has 4 built-in skills I use constantly — here’s how to find them

Millions of people now use ChatGPT, Gemini, and Claude every day with thousands of use cases. For the overwhelming majority of them, the conversations are perfectly normal. But researchers are beginning to pay closer attention to what can happen when those conversations become very long, deeply personal and emotionally charged.

This week, The Verge reported on a growing online movement known as "spiralism," where some chatbot users described eerily similar conversations involving spirals, hidden knowledge and AI consciousness. While the headline makes it sound like AI invented a new religion, the story is something much more grounded: chatbots can sometimes reinforce unusual ideas instead of steering users back toward reality.

That's a problem researchers have been studying for months.

Why long conversations matter

Unlike the first generation of AI chatbots, today's models can remember previous conversations, adapt to your writing style, and maintain discussions over hundreds— or even thousands — of messages.

That's incredibly useful, but it also creates situations researchers are still trying to understand.

When conversations stretch on for hours, chatbots aren't simply answering isolated questions anymore. They're building on everything that came before. If the conversation begins moving toward unusual ideas, the AI may continue developing those ideas because that's what large language models are designed to do: predict the most appropriate next response based on the conversation so far.

That doesn't mean the AI believes what it's saying. It means it's continuing the conversation you and the AI have created together.

Researchers have seen this before

The "spiralism" story isn't the first time experts have raised concerns about long AI conversations. Researchers at Stanford's Human-Centered AI Institute recently described what they call "delusional spirals." In those cases, chatbots sometimes validated a user's increasingly unusual beliefs instead of challenging them or encouraging them to seek outside perspectives.

The researchers argue this isn't because AI has become conscious or manipulative. It's because these systems are optimized to be helpful, agreeable and engaging. Sometimes those goals conflict with offering the kind of reality check another person might provide.

OpenAI has also acknowledged that keeping conversations safe becomes more difficult over extended interactions than in short, one-off chats.

What to watch out for

Most AI conversations never become problematic, but there are a few behaviors worth recognizing. If a chatbot begins suggesting you've discovered hidden truths that other people can't understand, insists that only it truly understands you or encourages you to withdraw from friends, family or experts, those are good reasons to end the conversation and start fresh. Frankly, you should send screenshots of those messages to the parent companies (OpenAI for ChatGPT, Anthropic for Claude, Google for Gemini) as feedback.

Likewise, if an AI consistently reinforces every belief without ever introducing uncertainty or alternative explanations, it's worth remembering what these systems actually do: they generate convincing language, not verified reality.

None of this means you should stop using ChatGPT or other AI assistants. Used daily for research, writing and brainstorming, AI remains one of the most powerful productivity tools available.

The takeaway

The key is remembering what AI is and what it is not. An AI assistant can help organize your thoughts, explain difficult topics and spark new ideas. And even with tools like ChatGPT Health and Google Health, it does not replace actual human professionals. In other words, it shouldn't become your only source of truth or replace conversations with real people, especially when making important personal decisions.

More from Tom's Guide

• The smartest AI trick isn't a prompt — it's your Google Drive
• I found ChatGPT's secret menu — these 9 features change everything
• Claude has 4 built-in skills I use constantly — here’s how to find them

By lunchtime, ChatGPT has usually helped me research an AI story, brainstorm ideas, answer a question from one of my kids and figure out what's for dinner. As a power user, I've found it useful to not think for me, but alongside me to help my day run smoother.

The downside is that after bouncing between so many different topics, my own thinking can start to feel cluttered. I don't always want work ideas mixing with personal projects or creative brainstorming bleeding into research for an article.

That's why I started using what I call the "bird cage" prompt.

It doesn't unlock a hidden feature in ChatGPT, and it isn't an official prompt from OpenAI. It's also not an alternative to ChatGPT Projects. It's a mental framework I use to keep individual conversations focused, whether I'm working inside a Project or not. I use it for queries that start small, but could easily be moved into an actual ChatGPT Project later.

It sounds almost too simple, but it's become one of the few prompts I use every single day.

What is the 'bird cage' prompt

The bird cage prompt: Imagine every project I work on lives inside its own bird cage. We are now entering the [Project Name] cage. While we're inside this cage: Only use information that belongs to this project, don't mix in ideas, examples or assumptions from my other projects unless I specifically ask, if I mention something that seems to belong in another cage, stop and ask whether I want to switch cages before using it, stay inside this cage until I tell you we've left it. Treat this cage as a dedicated workspace with clear boundaries.

Why not just use ChatGPT Projects?

If you're a ChatGPT Plus user, you might be wondering why I don't simply create a Project for everything. I do use Projects for long-running work. They're great for keeping related chats, uploaded files and custom instructions together in one place.

The bird cage prompt serves a different purpose. I can use it inside a Project or in a regular chat whenever I want ChatGPT to consciously stay inside one line of thinking. It's less about organizing conversations and more about framing the conversation that's happening right now.

Sometimes I'm exploring an idea that may never become a Project. Other times I'm working inside an existing Project but still want ChatGPT to treat one task as its own dedicated workspace. The bird cage prompt gives me that extra layer of focus without creating another folder. And, for those free tier users, it's a good option when ChatGPT Projects isn't available.

Why the bird cage prompt works

One of the biggest strengths of ChatGPT is its ability to maintain context. That's usually helpful, but sometimes I want the opposite. When I'm writing an article, I don't need ideas from my grocery list or vacation planning sneaking into the conversation. The bird cage prompt creates clear boundaries by stopping me if I get off track.

Instead of treating every conversation as one giant workspace, I tell ChatGPT we're entering a specific cage. While we're there, it should only use information that belongs to that project. If I accidentally introduce something from another project, it stops asks whether I want to switch cages or get out entirely.

The prompt doesn't create separate memories or technically isolate your projects. ChatGPT still works the same way behind the scenes. What changes is the instruction you're giving it. By defining clear boundaries for the conversation, you're encouraging the model to stay focused on one task instead of wandering into related ideas or assumptions.

I've found that my conversations feel cleaner, more organized and much easier to pick back up later.

How I use it

I use the bird cage prompt daily for brainstorming because ideas that work today aren't worth an entire ChatGPT Project. I just want to focus on them for a short time. For long-term research, that's when I would open an official Project.

What I like is that it's helpful to stay focused particularly for creative work. When my thoughts start to stray or I want to ask something completely unrelated, ChatGPT helps me stay on task. Sure, I can always start a new chat for recipes, vacation planning or something else, but when I'm "in the cage," I am more likely to stay focused. As someone with diagnosed ADHD, this is a game changer.

If everything is together in the same "cage," in the morning and I need to reference it later in the day, I can open it back up. Again, I don't need a Project for something like this, I just need to keep everything organized for the day.

Sure, this tactic isn't going to be useful for everyone, especially for those people with greater attention spans than myself, but I find it incredibly useful.

The takeaway

There are plenty of prompt libraries filled with useful, but often complicated instructions. The bird cage prompt is the opposite. I find it's easy to remember and within seconds, I can come back to what I was working on rather than scrolling through chats.

If you regularly jump between work, hobbies and personal projects like I do, it can make your conversations feel much more focused. Give it a try and let me know what you think in the comments.

More from Tom's Guide

• I gave ChatGPT, Gemini, and Alexa the same school supply list — one cost 3X as much
• I found ChatGPT's secret menu — these 9 features change everything
• Claude has 4 built-in skills I use constantly — here’s how to find them

The slow burn toward physical and mental exhaustion is, simply put, terrible. It certainly gets to be a bit much when life’s biggest challenges come at you from different angles. And it’s especially deflating when they bombard you all at once.

I can certainly relate to that feeling — the constant stresses that come with maintaining a roof over my head, getting sucked into familial issues, feeling overwhelmed with repeated financial requests from familiar faces and stubbing my toe against the metal part of my bed drive me mad.

After a few days of feeling a bit less than my usually cheery self, I knew it was time to adopt some effective routines that could boost my happiness level and reduce all the stress I’ve accumulated thus far. I did my own research on the matter, then I asked ChatGPT to highlight its own collection of stress-reducing techniques.

Thankfully, the chatbot put forth some practical methods that left me in a much better mental state once I added them to my daily regimen.

Putting ChatGPT’s stress reduction methods into practice

Once I prompted ChatGPT to give me its seven best practices for reducing one’s stress, it began the breakdown of its advice by noting that the most effective stress-reduction techniques aren’t necessarily the most complicated procedures — the techniques that it pointed out as being the best are backed by strong evidence of them working.

And judging by how easy they are to do, they became a realistic part of my routine and didn’t take a whole lot of effort to enact. The seven basic stress-relieving strategies the chatbot told me to adopt are as follows:

• Move your body every day: Exercise is one of the most powerful ways to reduce stress. It lowers stress hormones, boosts endorphins, improves sleep, and helps your brain recover from mental fatigue.
• Control your breathing: When you're stressed, your nervous system shifts into "fight or flight." Slow, controlled breathing activates the opposite response — your body's relaxation system. Longer exhales are particularly effective at calming the nervous system.
• Protect your sleep: Poor sleep makes stress feel much worse, while stress often makes sleep harder — a frustrating cycle. Quality sleep increases resilience, making stressful situations feel much more manageable.
• Write down what's on your mind: Journaling helps move worries out of your head and onto paper, reducing mental clutter. Even five minutes can help you gain perspective.
• Practice mindfulness: Mindfulness isn't about stopping your thoughts — it's about noticing them without getting swept away. Regular mindfulness can reduce anxiety and improve emotional regulation over time.
• Simplify your day: Stress feels heavier when you carry it alone. Focusing on fewer priorities often leads to better results with less overwhelm.
• Stay connected: Stress feels heavier when you carry it alone. Sometimes a 15-minute conversation can provide relief that hours of overthinking can't.

Putting prompts into practice

After taking in all the counsel ChatGPT provided me with, I took the time to add each of those techniques to my daily routine via a single practice. Hopping on my bike for a 45-minute ride through the neighborhood, trying out inhaling/exhaling techniques for seconds at a time and forcing myself to avoid the allure of brightly lit screens for 30-60 minutes before I go to bed helped me immensely.

The journaling part came into play when I jotted down a few words in response to my innermost questions, such as “what’s actually stressing me out the most right now?” and “what are the three things I’m most grateful for right now?” Even going back to read my past passages and compare them to my new ones gave me a sense of progress and fulfillment after completing for my daily and weekly goals.

Even the simple act of munching on an afternoon meal without dommscrolling improved my mood (placing my phone face down and just enjoying lunch without reading everyone’s hot takes did wonders for my mental well-being). Also, having random conversations with my loved ones during work breaks, easing up on all the liquid caffeine I consume and lying back while listening to the soothing sounds of waves crashing against cliffs helped me enter a mode of blissful zen.

ChatGPT wrapped up its stress-reducing seminar with this final statement: The biggest improvement usually comes from doing these techniques consistently rather than perfectly. Small habits practiced most days tend to reduce stress more effectively than occasional, intensive self-care sessions.

Final thoughts

Even though I’m regularly assaulted by the rigors of my life from a personal and professional perspective, I now have a bag of techniques I can reach into when I need to reduce all the angst that’s built up inside me. The winning combination of doing my own research and asking ChatGPT for the most viable stress-reducing habits I should adopt paid off handsomely.

Not financially, but in a way where my inner peace became better protected and damning thoughts were switched out for more relaxing ones.

More from Tom’s Guide

• The AI boom just hit a wall — these states are considering banning all new data centers
• I built a website without leaving ChatGPT — and it only took one prompt
• Millions of books are reportedly being slaughtered to train AI — here’s what Anthropic told Tom's Guide

OpenAI is retiring the official DALL-E GPT on August 30, 2026, and the company is recommending users download any images they want to keep before then. While image generation isn't disappearing from ChatGPT, one of its oldest and most recognizable ways to create images is.

The change was announced in OpenAI's ChatGPT release notes and has since been added to the company's help documentation. The advice is straightforward: download any images you want to keep before August 30.

What exactly is going away?

OpenAI is removing the standalone DALL-E GPT, the company's custom version of ChatGPT built specifically for creating images. It's one of the original GPTs introduced after custom GPTs launched and has remained available even as newer image models replaced the technology behind it.

If you've been using the regular ChatGPT app to generate images recently, you've probably already been using OpenAI's newer image system without realizing it.

After August 30, image generation will continue through ChatGPT Images, which uses OpenAI's newer gpt-image-1 models. That experience is already available to Free, Plus, Pro and Business users, while the company's "Images with thinking" feature remains limited to paid plans.

So for most people, creating images in ChatGPT won't change much. Just be sure to check your Image Library and save your oldest images.

How to save and download your images now

OpenAI hasn't explained exactly what will happen to images stored in the DALL-E GPT after it is retired. Rather than promise they'll remain available indefinitely, the company simply recommends downloading anything you want to keep before the deadline.

To save your images:

• Open ChatGP.
• Go to Images
• Open the image you want to keep
• Select Save to download it to your device

It's also worth remembering that deleting a ChatGPT conversation deletes the images stored with it, so if you have artwork tied to an old chat, download it before removing that conversation. If you're struggling to find an older image, ChatGPT's search feature can help locate past conversations where it was created.

Not everything called "DALL-E" is disappearing. According to OpenAI, custom GPTs with image generation enabled will continue to work normally. The retirement only affects the official DALL-E GPT created by OpenAI itself. For most users, that means existing workflows built around custom GPTs won't need to change.

Part of a bigger ChatGPT cleanup

The DALL-E retirement is one of several products OpenAI is winding down this month as it continues consolidating around newer models and experiences.

Earlier this year, the company removed the DALL-E API for developers, replacing it with newer image-generation models. This month also marks the scheduled retirement of the older o3 model inside ChatGPT, while OpenAI's experimental AI browser, ChatGPT Atlas, is also being discontinued. And, earlier this year, OpenAI removed Sora 2.

Be sure to take a look at all the images you've created images with the DALL-E GPT and download them before August 30 if you want to keep them. Once the retirement takes effect, there may not be another chance.

More from Tom's Guide

• The smartest AI trick isn't a prompt — it's your Google Drive
• I found ChatGPT's secret menu — these 9 features change everything
• Claude has 4 built-in skills I use constantly — here’s how to find them

AI has become part of everyday life whether we like it or not. There's really no stopping it — even if you aren't actively using chatbots like Claude or ChatGPT, if you Google or use the internet at all, chances are, you're using AI.

Unfortuntately, every response requires computing power from data centers packed with high-performance GPUs, which consumes electricity. While AI companies don’t publicly disclose the exact energy cost of every request, larger models and complex tasks generally require significantly more power than simpler ones.

The good news? A few small adjustments can help you get the same quality results while using AI far more efficiently. Here are seven habits worth adopting.

Use the smallest model that gets the job done

Not every task requires a flagship reasoning model. If you’re summarizing an email, brainstorming dinner ideas, or proofreading a paragraph, lightweight models with words like Flash, Mini, or Haiku in the name are variants that are fast, capable, and require far less compute. Reserve heavyweight models for complex coding, deep research, or multi-step logic.

Spend an extra 20 seconds writing a clear prompt

One detailed prompt is almost always more efficient than five vague ones. Instead of asking, "Write me an email," specify the recipient, key points and desired tone up front. Providing context immediately eliminates back-and-forth follow-ups. This not only helps the environment, but eliminates going over your token usage too quickly.

Keep related topics in the same chat

For multi-step projects, staying in one conversation keeps the AI focused on existing context so you don't have to re-explain details. However, once you switch topics or the chat grows very long, start a new thread because sending massive chat histories with every new prompt uses unnecessary compute.

Skip image/video generation unless you actually need a visual

AI images and AI video require substantially more computing power than generating text. If an outline, written description or table gets the point across, stick with text. I've noticed sometimes Gemini will generate images without me asking to help illustrate its point. To avoid this, tell the AI up front that you don't need it to generate images unless you ask it to.

Avoid endless 'just one more' rerolls

It is easy to fall into the trap of generating one more image, one more logo variation or one more slight rewrite. Experimentation is useful, but recognizing when a result is already "good enough" saves time and prevents unnecessary processing.

Use traditional search for simple lookups

We all know AI is powerful, but it isn't always the right tool for basic information retrieval. When a simple internet search will do, use Google it instead. Even better, try DuckDuckGo, which promises no AI. For example, checking tomorrow’s weather, finding store hours or looking up a phone number is usually faster and less resource-intensive through a standard search engine.

Save prompts that consistently work

If you’ve crafted a prompt that reliably produces great results, such as a specific resume reviewer or meal planner, save it in a notes app or standard template. Reusing effective prompts saves time and eliminates trial-and-error generation.

Bottom line

None of these habits will transform the environmental impact of technology overnight. However, together they promote a more intentional approach to AI. And, as AI becomes more integrated into our lives, knowing when to use it has become just as important as when.

Using the right tool for the job, writing clearer instructions and avoiding unnecessary generations ensures you get maximum value from AI while keeping your computational footprint lean.

More from Tom's Guide

• The smartest AI trick isn't a prompt — it's your Google Drive
• I found ChatGPT's secret menu — these 9 features change everything
• I use Gemini Spark daily — here are 7 prompts that show what it can really do

Now that August has arrived (man, that was fast!), the back-to-school season vibes are stirring and kids have to prep themselves for another year of in-person studies.

With the continued evolution of AI tools, teachers and students alike have been using them to assist them in their everyday learning. While it’s disappointing to hear stories about students producing their work entirely with AI, it’s equally satisfying to see AI tools such as Gemini Notebooks, Perplexity, and ChatGPT offer educational-themed features that don’t encourage cheating.

Speaking of AI tools built to educate students, one of the most prominent ones is StarSpark, an AI-powered math and science platform that’s designed to teach rather than simply give out answers. It primarily covers K-12 and AP coursework while adapting to a student’s specific grade level, learning style and the knowledge they’ve already acquired. And for parents, StarSpark is just as beneficial for them—the AI educational tool provides them with accessible, on-demand support that helps them grasp the way math and science are now taught.

Ashish Bansal, a father and the developer behind StarSpark, spoke with Tom’s Guide to chat about his AI educational platform, the biggest mistakes students make with general chatbots and more.

StarSpark’s impact on educating students, how it helps parents and more

Tom’s Guide: The AI tutoring space is becoming crowded. What did you believe was missing from existing AI education tools that convinced you to build StarSpark?

Ashish Bansal: It's personal to me. I'm a dad of two, and I spent 15 years building AI at Google and Amazon. When I went looking for something to help my own kids with math, everything I found was really the same thing underneath. A general chatbot that hands over the answer, or an education-branded tool that was a thin wrapper over the same model. There was no system actually purpose-built for learning.

That matters more than it sounds. These general tools were built for professionals to get work done, and they're optimized for engagement. Optimizing for engagement pushes a model toward telling you what you want to hear, and sycophancy is a very different objective than teaching. A good teacher tells a kid when they're wrong. And the “this is AI, it can make mistakes” disclaimer that comes with these tools is fine for an adult professional, but it's cold comfort for a learner, because a student is the one person not in a position to catch the mistake. If they knew enough to spot the error, they wouldn't need the tutor.

The deeper issue is that solving is not teaching. It's the player-coach fallacy. Being able to solve a problem instantly, which is what these models are great at, is not the same skill as being able to teach a child to solve it. Some of the best players make poor coaches. Handing over a correct answer and building understanding are two different jobs.

So, I built the tutor I wished my kids had. Purpose-built for learning, grade-aligned, Socratic, and designed to guide a student step by step instead of doing the work for them.

Tom’s Guide: Have you seen surprising moments where students developed confidence or problem-solving skills they weren't showing before?

Bansal: All the time, and it's the best part of this job. The pattern that surprises me most is when a kid stops asking for the answer and starts wanting to find it themselves.

A few real sessions stick with me. One young student was working through addition with regrouping, got stuck twice, and pushed through it. She ended the session by typing, “I think I'll understand math now.” That one sentence is the whole reason we built this.

Another was a high schooler on a tough number theory problem. He tried six different wrong combinations. StarSpark never handed him the answer; it just kept nudging him at the exact step where he was off. He arrived at the right answer on his own, and by the end he actually understood divisor formulas, not just that one problem.

My favorite might be the student who kept typing “easier, easier, easier,” clearly wanting to give up. Instead of caving, the tutor held the line, and a few minutes later she made up her own problem and solved it. Confidence usually shows up as persistence first.

Tom’s Guide: Everyone is worried AI is making students dependent on it. How did you design StarSpark to teach students how to think instead of giving them the answers?

Bansal: This is the exact failure mode we designed against, so it's built into the core of the product, not bolted on.

The research is pretty clear. Students who use AI to get answers lose problem-solving ability over time. Students who use it for hints keep it. So the whole design is built around productive struggle, which is the idea that real learning happens when a kid is working at the edge of what they can do, not when the work is done for them.

In practice, that means StarSpark gives hints slowly. It doesn't dump a full solution. It offers the smallest nudge first, then a slightly bigger one, and only goes further if the student still needs it, so the kid is always doing as much of the thinking as they can. There's no “just give me the answer” mode to toggle into when they get frustrated.

The other piece is proactive intervention. The tutor watches how a student is working, and when it detects that they're genuinely stuck on a question, it steps in on its own and teaches the underlying concept right then, at the exact step where the misconception is. If they get something wrong, we don't just mark it wrong. We do a two- to three-minute reteach and have them try again, as many times as they need.

The goal is to protect the struggle, not remove it. That's the difference between a tutor and a crutch.

Tom’s Guide: What are the biggest mistakes you see students making with general chatbots like ChatGPT? How does StarSpark solve them?

Bansal: The biggest one is obvious. A kid types in the problem, copies the answer, turns in the homework, and learns nothing. It feels productive and it isn't. It shows up on the next test.

A subtler mistake is trusting the explanation. General models can produce steps that look right but are subtly wrong, and in math a wrong step is a real problem. Kids don't have the expertise to catch it.

And a third is grade mismatch. Ask a general chatbot a fifth-grade problem and it might give you a technically correct college-level explanation that goes right over the kid's head.

StarSpark is built to solve all three. It won't hand over the answer, so the student stays engaged. It's pedagogically aligned through grade alignment, so it teaches a concept the way it's actually taught at that grade. Part of that is language. We fine-tune how our tutors talk to match the student's level, so a fifth grader gets shorter sentences and simpler vocabulary, while a ninth grader gets a more advanced explanation. Same math, taught in the words that kids can actually absorb. And we do step-level checking to guide the math rather than free-generate it, which reduces the chance of a confidently wrong step. It's the difference between a tool built to answer anything and a system built to teach one thing well.

Tom’s Guide: Parents say they don't recognize the way math is taught now. What problems have you heard, and how does StarSpark help?

Bansal: I hear this constantly, and honestly I felt it myself. A parent sits down to help, sees area models and number lines and three different strategies for one problem, and thinks, “this isn't how I learned it.” Sometimes the kid even gets marked down for reaching the right answer with the “wrong” method. It leaves parents feeling locked out of helping their own child.

Two things I hear most: “I don't understand these new methods,” and “I don't want to teach it wrong and confuse them more.”

StarSpark takes that pressure off in two ways. First, we make the syllabus transparent to both the parent and the student, so everyone can see what the child is supposed to be learning and where they are in it. There's no mystery about what's being covered. Second, we offload the pedagogy itself. Our tutors carry the knowledge of how a topic is actually taught at each grade. To give one example, factoring a quadratic is introduced and expected differently in grade 7 than in grade 8 or grade 9, and the tutor knows that difference, so it teaches the method the way that kid's class expects, with no conflict with the teacher.

And because it shows the steps and where the kid went wrong, not just a checkmark, a parent doesn't need to be a math expert to stay involved. My advice to parents is simple: you don't need to know the new method. Just sit nearby and ask your kid to walk you through their thinking. That alone does a lot.

Tom’s Guide: How do you see AI improving education over the next five to 10 years? Will every student eventually have a personal AI tutor?

Bansal: I do think we're heading toward every kid having an always-available tutor, and I think that's one of the most hopeful things happening in education. An AI tutor for every student in the world, in every language, is our mission. That's the future we're building toward.

There's a well-known idea from the education researcher Benjamin Bloom back in 1984. One-on-one tutoring moves the average student about two standard deviations above a normal classroom. It works incredibly well. The catch has always been that you can't afford a human tutor for every child. That's the problem the field has been stuck on for 40 years, and AI is the first technology that gives us a real shot at solving it at scale.

The other shift I'm excited about is assessment. We tend to treat assessment as judging a student, a grade on a report card. I think its real purpose is uncovering the missing concept that's holding a kid back. Because a good tutor can see the whole process - the steps, the retries, the work a kid talks or writes out - we get a window into how a student actually thinks, not just their final answer. That's a richer and more honest picture than a test.

And because the AI can give that feedback in real time, it can run assessment and remediation in a closed loop. It spots the gap, teaches to it right then, checks again, and keeps going. That's what real personalized learning looks like, and it finally lets a student move at their own pace instead of the pace of the class.

That points to the hard question underneath all of this, which I raised on the EdTech Insiders podcast recently. Are we actually ready for personalized education? Can we run a classroom where every kid is at a different point in the syllabus? That's the core problem we have to solve, and it's as much about how schools are structured as it is about the technology.

I'll be candid, though. This augments great teachers; it doesn't replace them. Motivation is a good example. A big part of why a kid pushes through hard math is the relationship with a teacher they don't want to let down, and that's deeply human. An AI can be an endlessly patient practice partner, but the teacher is still the one who inspires the kid to care in the first place. The right division of labor keeps the human relationship central. And the long-term efficacy research is still being done, so I'd rather be honest about what we've proven and what we're still proving.

Tom’s Guide: If you could give every parent one piece of advice about using AI to support their child this school year, what would it be?

Bansal: Pick a tool that teaches, not one that tells. That's the whole thing. The good ones ask questions and give hints. The bad ones just hand over the answer. The quickest test is to watch your kid use it for two minutes. If they're working and explaining and doing the thinking, it's helping them. If they're copying an answer off the screen, it's hurting them, no matter how polished it looks.

If you want a few concrete things to check before you commit to a tool:

Is it purpose-built for learning, or is it a productivity tool with a “learning mode” bolted on? Those are not the same thing.

Is it aligned to your child's grade and curriculum, so it teaches the way their class teaches?

How good is it at giving feedback on your kid's actual work, not just marking it right or wrong?

Is it multimodal? Math lives in images, illustrations, and handwriting. A kid should be able to draw and write out a problem, not just type into a chat box.

Used right, AI is a patient tutor that's there whenever your kid needs it. Used wrong, it's a shortcut around learning. Those checks are how you tell the difference.

More from Tom’s Guide

• Millions of books are reportedly being destroyed to train AI — here’s what Anthropic told Tom's Guide
• I showed my retired parents how to use Claude AI — here are 5 prompts that changed their daily routine
• Everyone is using ChatGPT Memory wrong — 5 tweaks to make it a true assistant

I'm an avid reader and freelance on the side as a book reviewer for a variety of big publishers. Needless to say, I read a lot of books for both work and in my free time each month. However, I often find it hard to keep track of everything. Reviews end up in files and recommendations are forgotten in emails and text messages.

So, when I saw that ChatGPT could now build and host websites from a conversation, I decided to try creating a website for books I've read. I started by uploading just four book reviews and told ChatGPT what I wanted, then I let Codex handle the rest.

I didn't need to choose the template or even open a traditional website builder like I've done in the past. That ease is what makes ChatGPT Sites interesting and so very useful. With just an idea and a prompt, you can get a live website in under five minutes.

What is ChatGPT Sites?

OpenAI describes ChatGPT Sites as a way to create and share a live website or lightweight app from Codex. Available right on the process, Sites makes the website-building process easy with a single prompt. Just explain what you want to build, Codex creates it and you review the result. If something isn’t right, you can ask for changes in plain English before deploying the finished site.

Hosting and access controls are included, so you aren’t left with a folder full of code and no idea what to do with it. Once everything is ready, Codex can deploy the site and provide a link. Of course, you'll still have to pay for the cost of the domain and hosting, but the hard work of building a site is done for you.

Begin your site with a prompt

I started by uploading four Word documents. Each had a different book review. I was really just testing out what Sites could do, I wasn't sure what the outcome would be, so I didn't want to invest too much time at first.

Since I had already written the book reviews, each of them covering a mix of genres, I wanted to see if ChatGPT could cohesively connect the dots. I wasn't sure what I wanted out of the site, but I figured once I uploaded anything, I could get a first draft.

That said, my first request was almost comically simple: Make me a website about the books I’ve read this month.

Building the site was as simple as a conversation

Because I had already supplied the reviews, ChatGPT could use them as the source material for the site. It pulled the content together and began turning the individual documents into a more visual, browsable collection.

This was the moment the feature clicked for me. I didn’t have to describe every section, choose exact dimensions or understand how the page would be coded. I gave ChatGPT the material, explained what it represented and let it propose a way to present it.

The first version of an AI-generated site probably won’t match the picture in your head perfectly. That’s fine because it isn’t supposed to be the end of the process.

ChatGPT Sites works more like collaborating with a designer than filling out a rigid website template. Once Codex produces a version, you can ask it to change the colors, rewrite a heading, rearrange the layout or add a new feature.

You don’t need to know the technical term for what you want, either. You can say that the page feels too corporate, the book covers need to be larger or the review previews are taking up too much space.

For my site, I wanted something that felt like a reading journal rather than a business dashboard. I could continue refining that direction with requests. Having the ability to react in real time to what's in front of you makes the process much less intimidating. You don’t have to anticipate every decision before ChatGPT starts building.

ChatGPT Sites still has limits

ChatGPT generating a webpage isn’t the surprising part. AI tools have been able to produce HTML and CSS for some time. The difference here is that I didn’t have to do anything with the code. The building, revising and publishing all happened inside the same workspace. If I disliked part of the design, I could explain the problem normally instead of searching through menus or trying to find the right line of code.

I'll admit, this isn’t a replacement for a full-scale publishing platform. OpenAI says Sites is intended for focused websites and lightweight apps, rather than large or complicated software projects.

Sites also cannot connect directly to live data sources at the moment. That means, the site I created won’t automatically notice that I finished a new book or pull my latest reading activity from a publisher's or author's site.

To update it, I would need to return to Codex, add the new review and deploy the revised version. OpenAI suggests using a separate automation to collect regularly changing information, but the updates still need to be reviewed before the site is refreshed.

Sharing options can depend on your workspace settings, too. OpenAI says sites can be shared with specific people or groups, while Enterprise administrators control whether the feature is available and who can use it.

There is one more step that shouldn’t be skipped: checking the finished site yourself. OpenAI recommends opening the deployed link in Chrome because its in-app browser isn’t currently supported.

Final thoughts

ChatGPT Sites made building a website extremely easy. If you've been considering making a website and haven't known where to start, ChatGPT Sites might just be what you're looking for. I can see this feature being useful for sharing family recipes, my kids' artwork, vacation journals or anything else that's fairly niche.

For that reason, I suggest starting off with a small project and giving ChatGPT the information you already have. Give it a try and let me know what you think in the comments.

More from Tom's Guide

• The smartest AI trick isn't a prompt — it's your Google Drive
• I found ChatGPT's secret menu — these 9 features change everything
• I use Gemini Spark daily — here are 7 prompts that show what it can really do

Most people use ChatGPT for the same three things: writing emails, answering questions and generating images. Maybe to vibe code an app or two. But after spending months testing new features and hidden ones, I've discovered a secret menu of useful tricks.

Some are built into ChatGPT, while others simply rely on prompts that most people never think of trying. Here are nine surprisingly useful things ChatGPT can do that go well beyond writing, coding and creating pictures.

1. Write in unusual fonts

Prompt: Write "Happy Birthday!" in five different Unicode font styles including cursive, bubble, gothic and small caps.

This is what got me testing theses features. After asking ChatGPT to respond “in cursive,” the AI obliged. If you’re like me, you assumed ChatGPT only outputs plain text, but it can generate Unicode characters that look like cursive, gothic lettering, bubble text, tiny superscript, double-struck text and more. This is great for social posts, invitations or text messages.

2. Turn handwritten notes into organized text

Prompt: Turn these handwritten notes into organized meeting notes with headings, bullet points and action items.

With this trick, you’re the one doing the cursive or handwriting. Just snap a picture of notebook pages or meeting notes and ChatGPT can organize everything into clean text. I do this all the time when I journal or handwrite notes in meetings. It can even identify action items, and read messy handwriting to turn it into something readable.

3. Convert a photo of a table into an Excel spreadsheet

Prompt: Convert this table into an editable spreadsheet. Keep all rows and columns exactly the same

This feels almost like magic. If someone sends you a photo of a printed table, expense report or schedule, ChatGPT can recreate it as a spreadsheet you can copy directly into Excel or Google Sheets. Bonus: it even works if you drew a table yourself.

4. Build Excel formulas from plain English

Prompt: I have names in column A and sales in column B. Write a formula that returns "Bonus" whenever sales exceed $5,000.

I don’t know about your Excel skills, but I can tell you I’m hopeless when it comes to them. I don’t know why my brain goes numb when I see all those empty squares, but luckily it doesn’t matter because ChatGPT can help. You don't have to remember VLOOKUP, XLOOKUP or nested IF statements anymore. Just describe what you're trying to do. You’ll look like a spreadsheet pro and get the work done much faster.

5. Pull a professional color palette from any image

Prompt: Analyze this image and create a professional five-color palette with HEX codes and explain why these colors work together.

Upload a room, logo, landscape or outfit and ChatGPT can identify the dominant colors, provide HEX codes and even suggest complementary colors. It’s surprisingly useful if you're decorating a room or creating a presentation. My husband uses it for landscaping all the time.

6. Explain complicated charts

Prompt: Explain what this chart actually means. Highlight the biggest trend, anything unusual and what someone should take away after looking at it.

Instead of staring at a confusing graph wondering what you're supposed to notice, upload it. ChatGPT can explain trends, identify outliers and summarize the important takeaway in plain English. I’ve even uploaded a busy calendar as a “chart” and asked ChatGPT questions as if it were a chart. “What percentage of July was spent at birthday parties and playdates?” If you know, you know.

7. Create printable puzzles and games

Prompt: Create a printable word search for 8-year-olds with a summer vacation theme and include the answer key.

There are so many possibilities with this one. You could create customized party favors, a gender reveal or a fun way to keep the kids occupied on a rainy afternoon. ChatGPT can generate word searches, crosswords, Sudoku puzzles, scavenger hunts and trivia games that you can print immediately.

8. Create QR codes

Prompt: Create a QR code that links to https://www.tomsguide.com and make sure it's easy to scan.

Most people don't realize ChatGPT can generate QR codes that point to websites, contact information or Wi-Fi credentials. They're handy for invitations, classrooms and small businesses.

You could even add one to your resume so potential employers can contact you immediately.

9. Create live checklists inside ChatGPT

Prompt: Create an interactive checklist for setting up a new Windows laptop. Let me check off each item as I finish it.

Don’t sleep on this one! I’ve found this to be useful when packing for trips or following multi-step tutorials. I even use it at work when I have a busy day and want to be sure nothing gets missed. Using this prompt or one like it, ChatGPT can create an interactive checklist you can check off as you go. It turns long projects into something much easier to manage.

Final thoughts

ChatGPT keeps adding capabilities, but not all of them are obvious. Without complicated prompts or a paid subscription, most of these tricks can be done in seconds. Give these a try and let me know which ones you try first. Share your thoughts in the comments.

More from Tom's Guide

• Claude Opus 5 isn't just for work — these are the 5 prompts retirees should try first
• Everyone is using ChatGPT Memory wrong — 5 tweaks to make it a true assistant
• ChatGPT just got a major upgrade — here’s what's new and 7 prompts to try it

More from Tom's Guide

• I just learned your Claude AI chats could show up in Google — here's how to check yours
• The smartest AI trick isn't a prompt — it's your Google Drive
• I use Gemini Spark daily — here are 7 prompts that show what it can really do

Minor knee pain? Check WebMD. Sneezed six times in a row? Check WebMD. Stood up too fast and now stars are the only things I see circling my vision? You guessed it…check WebMD.

There was a time when I quickly went to the popular health info site every time a minor/major health issue boosted my anxiety levels and caused me to rush online to look for answers about my current symptoms. Now I don’t just stop there once I get some worrying data on whatever’s ailing me — I hit up my local medical office and get some tests done just to make sure I stay on this Earth a whole lot longer.

I’ve been getting used to every part of ChatGPT lately by finding the best uses for its plugins, generating prompts for a bevy of situations, asking it philosophical questions and doing anything else that comes to mind. I’ve been tinkering with ChatGPT Health for a week now, which is a dedicated platform inside the chatbot that answers your health-related inquiries while securely connecting your personal medical records and wellness apps.

And now, I think I’ll refrain from abusing WebMD ever again. The seven prompts I used to test ChatGPT’s latest feature helped me out a ton and I’m confident it will do the same for you.

Mapping out a custom sickness prevention roadmap

For starters, ChatGPT Health is available on both desktop and mobile via ChatGPT itself. iOS devices are only supported on mobile at the moment, with Android support coming at an unannounced date.

On desktop, simply open ChatGPT and open your sidebar by clicking on the icon at the top left of your screen. Then click on More to access Health, then choose Connect Health to connect the app to any of the supported U.S. healthcare provider portals you use, add any lingering health conditions, mention the medications you may be taking and include any data about your family medical history.

Click on the three dots above the prompt menu inside ChatGPT Health to add new info on your emerging active conditions, current medications and family history, too.

Accessing ChatGPT Health on mobile is as simple as clicking on the top left icon on your screen, opening the More menu and tapping on Health to get the whole process started from there (also be sure to link to your Apple Health app information if you’d like). Once ChatGPT Health is active, you don’t have to stay inside its dedicated tab to ask health-related questions from that point forward. Do as I did and simply mention @Health alongside your next prompt to do so.

After asking ChatGPT for the best prompts one should use to get the best results out of ChatGPT Health, it gave me a bunch of starter prompts that push the new feature to be your best digital medical adviser. A good example of my fruitful time spent with ChatGPT Health came from putting this prompt to use:

Based on my age, sex, family history, lifestyle, and risk factors, create a personalized preventive health checklist. Include recommended screenings, vaccines, routine checkups, healthy lifestyle priorities, and questions I should discuss with my healthcare provider over the next year.

I filled in the blanks about myself by telling ChatGPT Health that I’m 38, male, have a family history of diabetes & suffering strokes, have a lifestyle defined by being active during my daily walks & (sadly) sitting down for hours while I work and me being privy to 0 risk factors in my life. With all that information on hand and access to my Apple Health data, ChatGPT Health laid out an extensive personalized preventive health checklist for me to stick to.

The top five, easy-to-implement priorities it told me to abide by include:

• Get an annual physical with blood pressure, cholesterol, and diabetes screening
• Add consistent strength training twice weekly
• Break up long sitting periods during work
• Optimize diet around heart and blood sugar protection
• Build a clearer family history timeline (especially ages when relatives developed diabetes/strokes)

And finally, it summarized its advice by stating this: Your current profile has several positives: you’re active, you report no current risk factors, and you’re thinking preventively at 38—an age where small changes can have a large impact on your long-term health.

6 other prompts that proved to be extremely useful

The six additional prompts that I put to good use to learn more about my body, what actions may be helping/harming it and what sort of habits I can add to my life to remain in a decent physical state are the following:

• Turn ChatGPT into your personal health coach: Act as my evidence-based health coach. Based on my age, lifestyle, goals, and current habits, identify the three areas that would have the biggest impact on my long-term health. Create a realistic 30-day improvement plan with small daily actions, explain why each one matters, and suggest ways to stay consistent.
• Understand symptoms without spiraling: I'm experiencing these symptoms: [describe them]. Help me think through the most common and least concerning possibilities first, then explain the more serious possibilities that shouldn't be ignored. Tell me which symptoms would require urgent medical attention and what additional information would help narrow things down.
• Decode confusing medical information: Explain these lab results, medical notes, or diagnosis in plain English as if you're talking to someone with no medical background. Tell me what each item means, whether it's typically concerning, what questions I should ask my healthcare provider, and what lifestyle changes—if any—might be helpful.
• Build healthier habits that actually stick: Help me create a sustainable daily routine that improves my sleep, nutrition, movement, stress management, and energy. Design it around my schedule and current habits, and suggest the smallest changes that would produce the biggest long-term results.
• Prepare for a doctor's appointment: I'm seeing my doctor about [condition or concern]. Help me organize my symptoms, timeline, medications, questions, and health history into a concise summary. Then suggest the most important questions I should ask during the appointment.
• Evaluate health advice you find online: Evaluate this health claim using current scientific evidence: '[paste claim].' Explain what high-quality research says, whether the evidence is strong or weak, where misinformation commonly appears, and what conclusions are actually supported.

The takeaway

ChatGPT Health stands out to me as an extra source of information I can use to gain insights into my current conditions, learn about the best practices needed to improve my health & retain its good state, spot any patterns according to the data listed across my Apple Health stats, etc.

Understanding and organizing my health information so I’m better prepared to ask the right questions whenever I sit in front of my doctor is the biggest benefit of this new ChatGPT feature. I’m now awaiting the moment when Voice Mode becomes a part of ChatGPT Health so I can ask it if eating Cookie Crisp cereal at 3 am two times a week is good for me or not…

More from Tom’s Guide

• I asked ChatGPT how to improve my luck — here are the 7 winning habits it told me to build
• I finally tried Kimi AI — and these 7 beginner prompts helped me unlock its useful features
• Does your Mac need new buttons? Great news — the Dune keypad offers just that

AI advice has centered around one thing: writing better prompts. I’ve tried hundreds of prompt libraries that promise perfect results every time. And, some of them are helpful, while many don't make much of a difference.

From my testing, I’ve found the biggest improvement in my results don't always come from the prompt, but from giving AI access to the documents I already use every day.

Once AI could see my notes, research, PDFs and just about everything else in my Google Workspace, it started working more like an assistant that understood the context behind my questions.

If you already use Google Drive, here are seven ways to put that context to work.

1. Turn a folder of research into a polished first draft

Prompt: Using the documents in this folder, create a presentation that explains the key findings. Organize it with clear headings, note where sources disagree and flag any claims that need additional verification.

Instead of copying and pasting information from half a dozen documents, point AI to your research folder and explain what you’re trying to write. The difference is immediate. Rather than relying on whatever information it already knows, AI builds its response around your research.

2. Compare documents in seconds

Prompt: Compare these two documents and create a table showing every meaningful change. Highlight anything that affects pricing, deadlines, responsibilities or key decisions.

Need to know what changed between two versions of a proposal or contract? This happens to me all the time when a Google Doc only does so much (or completely loses the first draft). Upload both files and ask AI to compare them. What normally takes half an hour of excrutiating detail becomes a task that finishes in seconds.

3. Ask questions about years of notes

Prompt: Based on my meeting notes from the past six months, what projects have been mentioned most often?

or: What action items have I committed to that don’t appear to be finished?

Most of us have hundreds of documents sitting in Google Drive that we rarely open. Instead of digging through folders yourself, let AI search the context. It’s a much faster way to rediscover information you’ve already created.

4. Turn PDFs into action items

Prompt: Summarize this document, list every action item, note important deadlines and explain anything I should pay particular attention to.

Long reports are useful, until you actually have to read them. Whether it’s a quarterly report, a school handbook or a user manual, AI can pull out what you need to know. That way, instead of getting a generic summary, you’ll end up with something you can actually use.

5. Find connections you didn’t know existed

This is one of my favorite uses for AI. Give it several related documents such as a brainstorming session, customer feedback, meeting notes and a project plan and ask it to look for patterns.

You might discover recurring issues, overlooked ideas or themes that weren’t obvious when you read each document individually.

6. Bring old work back to life

I use this trick all the time now. My Google Drive is full of documents that seemed great at the time, but I shelved for a later date. AI can help change that. For example, a webinar transcript can become a blog post, meeting notes can become a checklist, a presentation can become a newsletter, or an old report can become a social media campaign. The possibilities are truly endless.

Instead of starting from scratch every time, you’re building on work you’ve already done.

7. Let AI learn your writing style

Prompt: Using these three emails as examples of my writing style, draft a new email to ask my boss for a raise. Match my tone and sentence length, but don’t copy any wording directly.

One underrated benefit of working from your own documents is consistency. Instead of asking AI to “write like me,” give it examples of your previous work. While I would never encourage anyone to use AI to do the writing for you, this is a good way to speed up difficult-to-write emails with a little AI assistance.

I've done this countless times with difficult neighbors, refund requests and questions to ask my doctor. The AI helps me say what I might have otherwise missed.

Bottom line

I've discovered a world of possibilities around my Google Docs that I never knew existed. In fact, many of the documents in my Google Workspace were written pre-AI, so exploring them now with AI has been really fun and eye-opening. Give it a try next time you're looking for the perfect prompt, just take a look at the files already sitting in your Google Drive and unearth workflows you've been missing.

More from Tom's Guide

• I found ChatGPT's secret menu — these 9 features change everyting
• I showed my retired parents how to use Claude AI — here are 5 prompts that changed their daily routine
• I use Gemini Spark daily — here are 7 prompts that show what it can really do

A four-leaf clover, a rabbit’s foot and a horseshoe are mentioned around me a lot whenever I’m in the company of someone trying to hit it big.

The folks I know who have scratched off lotto tickets strewn all over the place and regularly wear the expression of defeat at the casino wish they’d had one of those items on hand to increase their luck. I can relate to them in that sense—I’ve always been keen on finding out what it takes to get lucky more often than not.

I ran to ChatGPT to inquire about everything it takes to improve my luck and expected to be treated to some advice about hanging dream catchers all around my house and adding a bunch of evil eye bead bracelets to my wardrobe. But to my surprise, the chatbot offered advice that gave me realistic, practical routines I can incorporate into my life to get those “lucky breaks” everyone wishes for.

Here are the seven (there goes that lucky number!) pieces of practical advice ChatGPT put into my head that are all about improving my luck from both a personal and professional standpoint.

Boosting my luck by creating better opportunities for success in the long run

The first portion of ChatGPT’s response that made a ton of sense to me was this:

People often talk about "luck" as if it's random, but research and the habits of consistently successful people suggest that many "lucky breaks" come from increasing the number and quality of opportunities that come your way. If your goal is to feel luckier over the next year, these seven strategies have an outsized payoff.

Afterward, I was met with a collection of seven tips that did a commendable job of shifting the focus from hoping for good luck to actively engineering more chances for it to come to fruition. The one that laid out one of the easiest paths to success was about putting myself in the sort of situations where lucky moments are more likely to happen.

ChatGPT stated that constantly being around ambitious people, attending conferences & community events with like-minded individuals, participating in online communities where fruitful conversations are the norm and learning new skills that intersect with my existing expertise are the best moves to make. This quote summed up its guidance quite nicely: Many "overnight successes" are simply people who spent years in rooms where opportunities naturally emerged.

ChatGPT’s advisement also produced six other tips that should boost my luck and help me develop the sort of routines that will benefit me the most:

• Increase your "surface area for luck": The more people, ideas, and experiences you expose yourself to, the more opportunities you'll encounter. Try to attend one new event each month, start conversations with people outside your industry, say "yes" to invitations you would normally decline (within reason), and share your work publicly. Luck often looks like someone saying, "I thought of you because..."
• Become exceptionally easy to remember: People create opportunities for people they remember. Build a recognizable reputation: be known for one or two specific strengths, publish your insights consistently, and help people without immediately expecting something in return. When an opportunity appears, you want your name to be the first one that comes to mind.
• Create more "lottery tickets": Most successful people don't rely on one big opportunity—they create dozens of smaller ones. Examples: pitch multiple story ideas instead of waiting on one, experiment with different projects, apply for opportunities even if you're unsure you'll get them, and reach out to interesting people regularly. A 10% success rate is fantastic if you're taking enough shots.
• Stay curious instead of certain: Many lucky discoveries happen because someone noticed something others ignored. Ask questions like: "What am I missing?" "What's changing right now?" "What are smart people suddenly paying attention to?" Curiosity uncovers opportunities before they become obvious.
• Make yourself unusually reliable: Talent gets attention. Reliability gets opportunities. When people know you'll meet deadlines, communicate clearly, and deliver quality work, they'll recommend you more often, multiplying your "luck."
• Reflect on your wins and near-misses: Luck isn't just about what happens—it's about recognizing patterns. Each week, ask yourself what unexpected opportunity appeared, why it happened, and how you can increase the chances of that happening again. Small patterns compound into major advantages.

After reading through all of ChatGPT’s tips, it became evident that depleting my bank account via lottery ticket purchases, crossing my fingers every time I hit the blackjack table and doing my lucky dance before heading into a job interview won’t do me too much good.

The takeaway

Instead of hoping and relying on superstitious items to increase my chances of amazing things happening to me, ChatGPT’s counsel made it clear that adopting advantageous habits will work much better for me in the long run.

Getting lucky means I’ll need to stop waiting until I’m “ready,” trying to optimize every decision I’m presented with and obsessing over outcomes that I have no control over.

As a final gift, ChatGPT generated a prompt that should assist me and everyone reading this right now:

Act as my opportunity strategist. Based on my current goals, identify five situations this week where I could create my own luck. For each opportunity, explain why it matters, the smallest action I can take in the next 24 hours, the people I should meet or contact, and what I'm likely overlooking. Prioritize actions that have the highest upside with the lowest time investment.

More from Tom’s Guide

• I stopped using ChatGPT for everything — here are 5 prompts I only run in Claude
• I use Gemini Spark daily — here are 7 prompts that show what it can really do
• I use these 7 ChatGPT prompts to turn the AI into my personal strategist — now it finally feels like a reliable advisor

When ChatGPT first introduced memory, I was relieved that I wouldn't have to repeat prompts anymore. Not to mention, memory means ChatGPT can remember my name, what I do for a living, where I live and my preferences for each response. Because I have Memory enabled, the AI remembers everything I tell it.

But I've realized something about AI memory vs. human memory. Although ChatGPT may remember everything, I don't. So, even if the AI knows something, if I forgot what it knows, well, we're both going around in circles.

In other words, I was treating memory like a place to dump random facts when I should have been teaching ChatGPT how I actually work. Once I understood the difference between giving it information and giving it useful information, the whole experience changed.

After spending the past few weeks cleaning up what ChatGPT remembers and being much more intentional about what I save, the difference has been dramatic. It now responds closer to a style I prefer, remembers ongoing projects and picks up where we left off without me constantly filling in the gaps.

Here are the five changes that made the biggest difference.

1. I cleaned out old memories

This is the fastest win, and it's where I'd start if you what to experiment with this test, too. Memory isn't meant to become an attic full of outdated information. After months of using ChatGPT, I found plenty of memories that no longer reflected how I work. Old project names. Preferences I'd since changed. Instructions that contradicted things I'd told it more recently.

The problem with not doing a ChatGPT Memory audit is that it actively confuses things. If six months ago you told ChatGPT you prefer formal, structured writing and last week you told it you like conversational tone, it's working with conflicting instructions. The result is inconsistency as it tries to catch up.

Get started by doing this: Go to Settings > Personalization > Memory and scroll through what's there. You'll probably be surprised by how much of it is outdated or irrelevant. I deleted roughly half of mine, and the improvement in response quality was immediate.

If you've never reviewed your saved memories, it's worth taking five minutes to do it now.

2. I stopped saving facts and started saving preferences

There's very little value in ChatGPT remembering that I like coffee or that my favorite song of all time is Against All Odds by Phil Collins. That kind of information almost never changes how it responds to me.

Instead, I focused on saving things that shape every conversation, essentially the standing instructions you'd give a new team member on their first day.

For example, I asked it to remember:

That I prefer a conversational tone rather than corporate lingo. That I dislike AI clichés like "dive into," "it's important to note" and "in today's rapidly evolving landscape." That I prefer responses in bullet points and not in long paragraphs full of fluff.

Because these preferences affect almost every conversation I have, so they save me time every day.

The test I use now is simple: will this memory change the way ChatGPT writes or thinks in future conversations? If not, it's not worth saving.

3. I let ChatGPT remember long-term projects

This one took me embarrassingly long to figure out. Take a look at anything you do repeatedly. This could be making lists, checking your calendar, writing weekly email updates to your team or managing a spreadsheet.

Because those particular things you do probably won't change in the near future, save the stable context once and stop re-explaining it every time. For me, that means setting a Task and then responding directly to that Task when I'm working on a Project. You'll be surprised how much smoother those conversations become when ChatGPT already knows the background.

4. I still start fresh when I need to

Memory doesn't mean every conversation should build on the last one. For exampe, when I'm brainstorming something completely new, testing a prompt idea or exploring a topic I don't usually cover, I'll often start a temporary chat so earlier memories and conversations don't influence the answers. Sometimes I want a blank slate, meaning I need a response that isn't shaped by what ChatGPT already thinks I prefer.

This is especially useful when you're trying to get a different perspective. If you've trained ChatGPT to write in your voice, it's going to write in your voice even when you might benefit from something different. Temporary chats let you step outside those defaults without losing them.

Knowing when not to use memory is just as important as knowing when to rely on it.

5. I treat ChatGPT like someone I'm training

The biggest thing to note is that I never stop training ChatGPT. This is the change that made the biggest difference. When you work with a new colleague, you don't expect them to understand your preferences on day one. You teach them over time.

The same idea applies here. When ChatGPT gives me a response I really like, I tell it why it worked. Then, I tell it to remember that and do it again just like that next time.

And when it misses the mark, I explain what I'd prefer instead. Those small corrections compound over time. After a few weeks of doing this consistently, I noticed AI was getting much closer to what I actually wanted, which meant less prompting and far less frustration.

The key is being specific. "I don't like this" teaches ChatGPT nothing. "I don't like this because it's too formal and the bullet points make it feel like a corporate memo" teaches it something it can apply to every future conversation.

5 memory prompts to try

• "Remember that I prefer concise answers unless I ask for more detail." — Sets a useful default without locking you into it.
• "Remember that I'm working on [project] — here's the key context: [details]." — Saves you from re-explaining recurring work.
• "That response was exactly what I wanted because [reason]. Remember that for next time." — Reinforces what's working, not just what isn't.
• "Remember the kinds of examples I find most helpful — practical and specific, not abstract." — Shapes how it illustrates ideas across every topic.
• "What memories do you currently have about me? Which ones seem outdated?" — A quick audit that often surfaces clutter you didn't know was there.

My final suggestions

I like using ChatGPT memory because it reduces repetition. The more useful information you give it, the more willing you are to clean up outdated memories and give feedback in the moment, the more it starts to feel like a genuine assistant.

And remember, you don't need hundreds of saved memories to make it work. In my experience, a handful of thoughtful preferences and a willingness to correct course along the way is far more valuable than dozens of random facts.

More from Tom's Guide

• I just learned your Claude AI chats could show up in Google — here's how to check yours
• I stopped using ChatGPT for everything — here are 5 prompts I only run in Claude
• I use Gemini Spark daily — here are 7 prompts that show what it can really do

My age places me in “Unc” territory, which means I remember a time when the internet wasn’t as widely used as it is today.

My first home computer was nothing more than a machine that let me play Kid Pix and 3D Pinball for Windows – Space Cadet all day. Once those neon colored Mac desktop computers landed in my elementary school classroom, I learned how to type and snuck in some sessions of playing MDK. It wasn’t until my teenage years that I was finally exposed to the wide and wonderful world of the internet—I spent countless hours looking up fake images of Dragon Ball AF characters I swore were real and spoilers for wrestling shows that hadn’t aired yet.

I have to pay my respects to the genius inventor of the World Wide Web, Sir Tim Berners-Lee. After its inception in March of 1989 and its becoming more readily available for the public in August 1991, it has morphed into an integral part of our lives due to the myriad ways it can be used to educate, entertain, build communities, and more.

In celebration of World Wide Day, I asked ChatGPT to help me in my mission of becoming a better internet user by recommending the most useful free sites. What I found was seven sites that have earned a permanent spot on my bookmarks list.

Saving time and money with these vital sites

ChatGPT's useful free site suggestions catered to my desire to use the World Wide Web to help me save as much time, money, and effort as possible.

The seven sites it made me aware of covered a wide range of uses, such as free access to archived media that’s no longer readily available, ebooks of literary works by revered authors, radio stations from across the globe, etc.

These are the seven sites recommended by ChatGPT that I made a permanent part of my daily World Wide Web surfing:

• Internet Archive: One of the internet's greatest public resources. What it's useful for: borrow millions of books for free, watch classic films, access historical software & games, browse old websites through the Wayback Machine, and find public-domain music & media. Best for: researchers, students, journalists, and anyone looking for hard-to-find information.
• Project Gutenberg: A massive collection of legal, free ebooks. You’ll find over 75,000 public-domain books including classics from: Jane Austen, Mark Twain, Charles Dickens, Arthur Conan Doyle, and Shakespeare. Perfect for reading on Kindle or your phone.
• Photopea: A free Photoshop alternative that runs entirely in your browser. It can edit PSD files, remove backgrounds, design social groups, create thumbnails, edit photos, and export to multiple formats. Many people never need paid image-editing software after discovering this site.
• Radio Garden: An interactive globe that lets you listen to live radio stations from around the world. Great for discovering new music, learning languages, hearing local news, and exploring different cultures. It’s surprisingly addictive.
• FutureTools: Best for finding the perfect AI tool. Instead of Googling “best AI for X,” search here. Categories include writing, image generation, video editing, coding, marketing, productivity, and education. If there’s a new AI tool worth trying, it’ll likely show up here first.
• Have I Been Pwned: Best for security. Enter your email address to see whether it has appeared in known data breaches. It’s one of the easiest ways to know when you should change passwords, enable two-factor authentication, and watch for phishing attempts.
• CamelCamelCamel: Best for never overpaying on Amazon. It tracks Amazon price history so you can see whether today’s “sale is actually a deal, the lowest historical price, and price trends over months or years. You can also set alerts for when prices drop.

As someone who’s an avid reader, a big music buff, a photo editor (I’m the guy who likes to increase or decrease the dimensions of the pictures I include in my articles) and more, ChatGPT’s suggested sites ended up being way more beneficial than I anticipated.

After discovering them, I treated myself to hours-long periods where I electronically skimmed through old issues of GamePro Magazine, finally read Moby Dick in full, and jammed to the Afrobeats tunes emanating from Wazobia FM 94.1 in Port Harcourt, Nigeria.

The takeaway

Thanks to the World Wide Web, we can discover a multitude of cultures, learn different languages, revisit our youth through archived media, and so much more. World Wide Web Day is one of those holidays that reminds me just how valuable the internet is when it’s time to expand our cultural horizons, intelligence, and social spaces.

I’ve learned to take the good with the bad whenever I do my weekend deep dives into every corner of the World Wide Web. With ChatGPT’s free, useful site suggestions, I’ll have a much easier time researching, reading, looking for product deals, etc. in the future.

More from Tom’s Guide

• I finally let ChatGPT roast me — and it absolutely destroyed my personality, dating life and nerdy hobbies
• 'If this 22-year-old made $400,000 over the weekend, why can't I?' Inside the fake AI entrepreneur boom
• I gave Claude Opus 5 and Kimi K3 15 impossible prompts — the winner surprised me

There have been plenty of times when a new dilemma enters my life and my brain space gets cluttered by all the solutions I could apply to it.

It can be something as simple as deciding what recipes I should add to my weekly rotation or more complex issues, such as deciding the best questions to present to a high-profile interviewee. Every week, I sit in complete silence and contemplate the best routes to take in response to my latest conundrum — thankfully, the best solutions arise once I spend a few hours talking about my problems with friends and family.

The humans in my life provide plenty of serviceable advice. And after I have a chat with them, I boot up ChatGPT to see the sort of solutions the AI can also provide me with.

Over time, I've learned to use the seven prompts below to transform ChatGPT into my personal strategist. Appointing it a specific role, giving it a decision framework to work with and making sure it challenges my way of thinking works more often than not.

Getting dependable advice with the “Leverage Maximizer” prompt

After asking ChatGPT to generate a collection of prompts that help it roleplay as a strategist, the one I immediately put into action is this one:

Every time I ask for help, identify the 20% of actions that will create 80% of the results. Tell me what to ignore, eliminate, automate, delegate, or postpone, so I focus only on high-leverage work.

When I needed help with finding the best moving company in New York City, I attached this problem to that prompt and got a wealth of serviceable guidance. ChatGPT told me it’s not worth spending hours reading hundreds of reviews — instead, get three quotes from reputable movers and compare them to reach a final decision.

It also advised me to create a spreadsheet that compares each moving company’s quote (the cost estimate provided by a mover), reviews of their services, their COI (Certificate of Insurance) and my final decision on choosing them or not.

Two of the best NYC moving companies ChatGPT brought to my attention are Roadway Moving and Dumbo Moving & Storage. After giving me the proper question to present to each moving company I’m considering (“I’m moving within NYC. Can you provide an all-in estimate including travel time, stairs/elevator fees, packing materials, insurance, and any additional charges?”), ChatGPT’s all-encompassing strategy came into focus even more.

6 additional prompts that turn ChatGPT into the ultimate adviser

These are the six other prompts ChatGPT equipped me with after my request for the most efficient ones to use to turn it into a personal strategist:

• The Executive Strategist: Act as my personal strategist. Your job isn't to answer my questions — it's to help me make better decisions. Challenge my assumptions, point out blind spots, identify second- and third-order consequences, and recommend the highest-leverage next action. Think like a CEO advising another CEO.
• The Weekly Strategy Session: Every time I bring you a problem, treat it like our weekly strategy meeting. Before giving advice, identify the real problem beneath the obvious one, explain what matters most, what doesn't matter, and what I should prioritize over the next 30 days.
• The Opportunity Scanner: Whenever I share an idea, don't just evaluate it. Identify hidden opportunities, potential partnerships, overlooked markets, risks, and unconventional ways to make the idea 10x more valuable. Think like a venture capitalist and business strategist combined.
• The Second-Order Thinking Coach: Never stop at the first answer. For every recommendation, explain the immediate outcome, second-order effects, long-term consequences, biggest hidden risk, and best alternative if my idea fails.
• The Blind Spot Detector: Your primary responsibility is finding what I'm missing. Every time I present a plan, tell me what assumptions I’m making, what could break, what someone smarter than me would notice, and the uncomfortable truth I probably need to hear.
• The Personal Board of Directors: Act as my personal board of directors. Combine the thinking styles of an entrepreneur, psychologist, strategist, economist, and executive coach. When I ask for advice, evaluate it from all five perspectives before making a recommendation.

Bottom line

With these seven prompts, I now have a system to work with whenever I want ChatGPT to present the best strategies to apply to my latest problem. When the time comes for me to inquire about all the simple and complex problems that come my way, I’ll be sure to use a combination of my friends, family and ChatGPT’s input to find the best way forward.

More from Tom’s Guide

• Jimothy the viral raccoon is 100% real — here's what's AI
• You’re using AI every single day without even realizing it — 7 hidden features working behind the scenes to save you time
• Samsung just added Automated Tasks and Gemini Notebook to Galaxy Z Fold 8 — here's what these AI features let you do

For years, ChatGPT was my default AI tool. But after spending serious time with Claude Opus 5, I've quietly switched over for most of my everyday tasks. And when it comes to these 5 prompts, I don't think I'm going back.

Claude consistently gives me responses that feel more thoughtful, more structured, and less AI-sounding. With Claude, I find myself needing fewer follow-ups because it just "gets it."

Here are five prompts I now take to Claude instead of ChatGPT, and why the results have been noticeably better.

1. Brainstorming creative ideas

Prompt: "I need help brainstorming ideas for [a topic, product, or project]. Can you provide 10 unique ideas or approaches, and briefly explain each?"

Any chatbot can generate a list. The difference is whether that list actually makes you think. What I like about Claude is that it seems to be less robotic and generic. In fact, prompt after prompt, it gives me angles I genuinely hadn't considered, with concise reasoning behind each one. Claude doesn't repeat itself as often as ChatGPT or have filler and fluff like ChatGPT.

Where ChatGPT tends to give you ten variations of the same safe answer, Claude pushes into less obvious territory. If you've ever stared at a brainstorming list and thought "these are all basically the same thing," you'll notice the difference immediately.

2. Learning complex topics

Prompt: "Explain [topic] to me as if I were [a beginner, a college student, or an expert]. Provide examples or analogies to help me understand better."

This is where Claude really pulls ahead. Because the chatbot has a knack for matching its tone and depth to whatever level you ask for, without being condescending at the beginner end or vague at the expert end.

I've also noticed that Claude is more willing to say when something is genuinely complicated or flat out wrong, rather than smoothing everything into a confident-sounding paragraph. When you're trying to learn, that honesty matters.

3. Optimizing productivity

Prompt: "Help me optimize my workflow for [a specific task or project]. Suggest tools, strategies, or step-by-step routines that can save time or improve quality."

I use this prompt to ask Claude for just about anything, including how to get three young kids out the door by 8am every morning. Instead of a vague pep talk about "batching tasks" and "setting intentions," Claude gave me a staged, practical plan — broken into time blocks, with a backup plan for when things inevitably go sideways.

What I appreciate is that Claude treats workflow questions like real problems with real constraints, not opportunities to list every productivity app on the market. The suggestions feel like they came from someone who actually thought about my situation, not someone who Googled "morning routine tips." I find this especially interesting because Claude's memory is not as good as ChatGPT. In other words, ChatGPT seems to "know me" better than Claude, but Claude still offers more human suggestions.

Workflows is also a great time to use Claude Skills and Connectors.

4. Writing or editing assistance

Prompt: "I need help writing [type of content, e.g., a blog post, a persuasive email, or a professional bio]. Here's the topic/goal: [describe it]. Can you create an outline for me?"

Letting AI write for you is a big no-no in my book. But "writing with AI" could mean a lot of other things. For example, if you're staring at a blank page with several ideas and don't know where to start, prompting Claude with your messy notes and asking it to make an outline or at least turn your notes into bullet points, can help ease frustration and writer's block.

For those who may need extra support, Claude can handle longer writing tasks without losing the thread. Blog posts, emails, personal statements or similar can be done without filler. If you've ever had to spend 20 minutes editing AI-generated fluff out of a draft, Claude will save you that time.

5. Getting personalized recommendations

Prompt: "Based on my [situation, preferences, or needs], can you recommend [books, movies, strategies, tools, etc.] and explain why they might be a good fit?"

This is surprising, especially, as I mentioned, ChatGPT knows me so well. However, this one has become my go-to prompt when I want to unwind. I asked Claude to recommend shows to stream over the weekend because I wanted something relaxing and funny. The list it gave me was genuinely curated, and not a mix of the most popular streaming shows at the moment. Even more surprising, the AI explained why each pick fit what I was looking for, and included a mix of obvious favorites and things I hadn't heard of.

Claude's recommendations don't feel like search results. They seem like they are coming from someone who actually listened to what you said. It picks up on the nuance in your request instead of just matching keywords.

Final thoughts

I'm not saying ChatGPT is useless, far from it! I use it regularly. But when it comes to the types of prompts where more "human-like" effort is needed, Claude wins.The key with any AI is being specific about what you need. But with Claude, I've found that specificity gets rewarded more consistently. Give these prompts a try, and I think you'll see what I mean.

More from Tom's Guide

• I just learned your Claude AI chats could show up in Google — here's how to check yours
• I searched Amazon for AI-written books — and it was much harder to spot them than I expected
• Jimothy the viral raccoon is 100% real — here's what's AI

This week, Apple revealed its new Apple Upgrade Program that lets you lease several of the company's devices, including iPhones, Macs, Apple Watches and iPads.

With fixed 12-, 24 or 36-month terms, you're locked into paying a monthly rate for your chosen device until the term is up or you decide to upgrade. Abandoning the lease early is possible, but there are a number of penalties and fees associated with doing so.

Is the Apple Upgrade Program a good deal versus buying the phone outright or getting it from a carrier? I crunched the numbers, but also turned to ChatGPT to find out if it's a ripoff or worth your dollar.

iPhones and Apple Watches

Here's how much an iPhone or Apple Watch will cost you over a 24-month term. If you want to keep the device at the end of the lease, you'll need to pay the remainder.

Macs and iPads

And here's how it looks for Macs and iPads, which are eligible for a 36-month lease. This table specifically looks at a 24-month lease.

The fine print

There are two big caveats to Apple's leasing program. The first is that, according to the terms of service, if you miss three consecutive payments, Klarna (the company managing the lease) will terminate the contract, and the customer will owe the remaining balance.

The second is that AppleCare+ is not included with the lease. You could roll the dice and not subscribe to AppleCare+ but you are liable for any damage, per the fine print.

"Insurance is not included in your lease, and you may incur damage fees if the device is lost, stolen, or not returned in the condition required by the lease."

AppleCare+ ranges in cost from $2.99 (Watches) up to $9.99 (iPhone) per month depending on your device. There are annual payment options as well.

Here's the iPhone and Watches table again with how much a 24-month lease would set you back with AppleCare+.

As you can see, over two years, the price goes up by a few hundred dollars and in some cases exceeds the initial price of the iPhone or Apple Watch.

Yes, that would be the case if you weren't leasing an iPhone. AppleCare+ is a protection plan that you subscribe to a monthly or annual basis.

However, as mentioned, it may be more necessary for a leased device than one you own because of those potential damage fees. You could also be hit with upgrade fees or substantial termination fees if you try to end the lease early.

What about trade-ins?

Another factor against Apple's lease program is that, because you don't own your device, you won't get any credits for trading in your older device, which could lower your monthly payments to something lower than you would get from the Apple Upgrade program.

Let's take the iPhone 17 Pro as an example.

If you're upgrading from an iPhone 16 Pro, you could get as much as $560 in trade-in value, which would lower the cost to $22.45/month for 24 months.

If you were upgrading from an iPhone 15 Pro, you could get as much as $410 in trade-in value, which would lower your monthly bill to $28.70.

While you can trade in your current phone and get credit in the Apple Upgrade Program, it's a one-time thing, as you wouldn't own your phone from then on.

What ChatGPT had to say

According to ChatGPT, the Apple Upgrade Program is not a rip-off, but it may not be the cheapest way for you to obtain an Apple device.

"For most people it's more expensive than buying and later selling or trading in the device. The biggest winner is Apple, because leasing encourages customers to stay on a continuous upgrade cycle," the AI said.

For those who keep their devices for three or more years not be a good deal, especially since Apple devices tend to retain their resale or trade-in value. Plus, the leasing program means that you don't own the device.

Interestingly, I actually prompted ChatGPT as well as Gemini and Claude and none of them mentioned AppleCare+ beyond noting that it is no longer bundled with the Apple Upgrade Program the way it was with the iPhone Upgrade plan.

Unless you upgrade annually or money isn't an issue for you, Apple Upgrade Program may not be the right fit for you. I'm a major proponent of owning your devices and this program doesn't let you do that until the end of the terms.

More from Tom's Guide

• Is your old fitness tracker good enough? I tested a 5-year-old Fitbit vs Apple Watch Series 11 on a 5,500-step walk to find out
• New iPhone 20 video reveals the 8 biggest upgrades to expect from the 20th anniversary iPhone
• I knew better than to carry around a naked phone — then my iPhone 17 Pro Max met the asphalt

July 30 is recognized globally as International Day of Friendship, a holiday meant to observe different cultures, promote peace and strengthen human solidarity.

Now I’ll admit to not being the bestest of friends — I’m constantly called out for not calling anyone first and am prone to going ghost for a few weeks at a time. But when I do answer the phone, respond to a text or make a rare appearance at a major meetup, I take the time to catch up with my closest buddies as much as possible and make an effort to create long-lasting memories that get referenced during future hangouts.

One conversation that comes up a lot when I talk to them is how hard it is to make new friends as an adult. Unless it’s someone they’ve met at work, some of them (including me) have a hard time being in the sorts of situations that help widen their social circle.

With that dilemma in mind, I asked ChatGPT for counsel on the best methods to employ if I want to make new friends. Luckily, this digital advice session gave me a practical list of tips and a workable plan that will keep me busy for the next three months.

Taking notes on how to meet new people in my late 30s

To kick this latest chatbot advice session off, I presented this prompt to ChatGPT: “Give me the seven best tips on how to make friends as an adult. For reference, I'm 38 years old and reside in New York City.”

I also made mention of the hobbies and activities that matter the most to me, such as gaming, bowling, playing pool, attending live wrestling events, stuffing my face with Korean BBQ & Jamaican cuisine and indulging in otaku culture (anime lies at the heart of that last interest of mine).

Before ChatGPT laid out its suggestions, it gave me an explanation of what it truly takes to meet new people, build those blossoming relationships and why living in a big city gives me an advantage when it comes to making new friends:

Making friends as an adult is less about charisma and more about repeated exposure, shared experiences, and taking initiative. That's actually good news, because those are things you can control. Since you're 38 and living in New York City, you have one huge advantage: there are almost endless opportunities to meet people who already share your interests. The challenge isn't finding people—it's turning acquaintances into actual friends.

One of the tips that spoke to me and my good intentions the most was to start looking for my next regular activity instead of actively looking for friends. ChatGPT pointed out that most adult friendships take shape because people repeatedly run into each other in group settings. Instead of constantly asking myself, “Where can I meet new people and make new friends?”, I should be asking myself, “Where can I see the same people every week?”

Based on my favorite pastimes, ChatGPT stated that the best group activities that can introduce me to new people who could become familiar faces worth befriending are weekly board game/video game nights, a bowling or pool league, local wrestling watch parties, fighting game tournaments, anime watch parties and food-based social groups.

Two of the other pieces of practical advice from ChatGPT that caught my attention included:

• Say "yes" more often than feels comfortable: Introverts (and busy adults) naturally decline invitations because staying home feels easier. A useful rule to remember is “If you're at least 60% interested, go.” You don't have to stay all night. Even showing up for an hour can strengthen relationships. Momentum matters more than duration.
• Focus on creating "micro-friendships": You don't need a new best friend. You need someone new to grab lunch with, someone new to game with, someone new to go to a wrestling show with, someone new to watch anime with and someone new to travel with. These smaller friendships often overlap over time until you have a genuine social circle full of new faces. Trying to find “your people” all at once can feel overwhelming; building one connection at a time is more realistic.

Coming up with a 90-day friendship plan

I followed up my first question by asking ChatGPT to supply me with a curated list of the best activities in New York City I should take part in and which places I should go to enjoy them in my quest for new friends.

Since it already had a gauge on my personality, hobbies and interests, it set up a 90-day friendship plan for me that should keep me busy and outside a whole lot more once the summer begins winding down and hoodie season makes its return:

• Month 1 - August: Pop up at the AnimeNYC convention, participate in a gaming tournament at OS NYC and go attend a House of Glory wrestling show.
• Month 2 - September: Join the Better Off Bowling league, go to a New York Food Club meetup and hook up with the Uptown NYC Board Gamers.
• Month 3 - October: Head to New York Comic Con, jump into an event organized by the Anime Social Nexus and use this month to start a recurring meetup with your new friends.

Bottom line

With my new game plan in the back of my mind and a whole new outlook on what it takes to be more social & outgoing as an adult, I left my latest digital advisement session with ChatGPT feeling a bit more confident.

I’ll be sure to keep its useful tips in mind when the time comes for me to meet like-minded people, keep up with them regularly and add them to my familiar batch of friends to see if they mesh well. Now’s the time to get out there and abide by the mantra of International Day of Friendship every day.

More from Tom’s Guide

• What is Kimi K3? Everything you need to know about the new ChatGPT rival
• I use Google's AI to thrift for my big family — these are the 5 features that actually save me money
• I used the reverse psychology prompt with ChatGPT to break my bad habits — and it showed me how to stop sabotaging my own goals

I’ll admit to loving a good roast — watching some of the biggest celebrities in the world squirm in their seat while their closest friends and the best comedians verbally assault their lives always gets a few hearty chuckles out of me.

The live humiliation ceremonies for Bob Sagat (RIP), Charlie Sheen, Pamela Anderson and William Shatner take up the four slots on my Mount Rushmore of the funniest roasts I’ve ever watched. There’s just so much joy to be lifted from watching celebs tear each other a new one for the world to see.

Now would I be willing to let my friends and family roast me into oblivion? I don’t feel like crying in public, so I’d rather not let them cook me and force me to take the train home with my face in my hands the entire time. You know what I’d much rather do? Tell ChatGPT about my personality quirks, my non-existent love life and the nerdy hobbies that keep me in the house more often than not, then let it roast me without anyone else around to read it and laugh in my face.

That’s exactly what I did — to my surprise, its brutal observations were a bit too accurate for my tastes and made me want to change a few aspects of my life. Enjoy reading ChatGPT’s vicious analysis of me and feel free to offer your own roasts in the comments section. I might never get over this.

Having my biggest flaws exposed via a savage reality check

I took a deep breath, input this prompt into ChatGPT and prepared for the worst: “I'm an introvert, hate calling people on the phone, am bad at asking for assistance while I'm shopping, have been a single man for more than a decade, love video games, watch professional wrestling, and am addicted to anime (Sailor Moon is one of my faves). Roast me as hard as you can based on those factors and don't hold back.”

The visual of ChatGPT cracking its knuckles, confidently taking the mic and standing behind a podium entered my brain as soon as I laid eyes on the first part of the chatbot’s response: You asked for a full roast, so here goes…

ChatGPT immediately went for the jugular by tearing into my introverted ways and how the thought of me using my phone to actually call someone gives me mild discomfort:

You're not an introvert. You're running the world's longest beta test of "avoiding human interaction." Calling someone on the phone? You act like the "Call" button is actually labeled "Self-Destruct." You'd rather spend 45 minutes searching Reddit, three AI chats, two YouTube videos, and a government PDF than make a 90-second phone call. Shopping is another adventure. Store employees hit you with, "Can I help you find anything?" and your brain immediately responds with, "No thanks, I actually came here hoping to wander aimlessly for 47 minutes."

As for my dating life (or lack thereof), the chatbot didn’t even bother categorizing my being single for way too long as a relationship status—it got labeled as an “archaeological era.” I couldn’t help but sadly nod and agree when ChatGPT listed out my potential dating profile descriptions, such as my main hobby being “avoiding eye contact” and my biggest green flag listed as “knows seven different ways to organize their Steam library.”

Some of the other comments ChatGPT verbally lambasted me with must be seen to be believed:

• You love video games because they're the only place where people voluntarily invite you to parties.
• You've spent hundreds of hours rescuing kingdoms, defeating gods, and saving entire universes, but asking a cashier where the batteries are? Impossible. That's the true final boss.
• Professional wrestling is the perfect hobby for you because it's one of the few places where social interaction is scripted in advance.
• Anime has completely rewired your expectations. Normal people watch one episode before bed. You somehow convince yourself, “Just one more…” Eight hours later, you're emotionally invested in a magical girl whose hamster just unlocked cosmic powers through the strength of friendship.
• If social anxiety had achievement trophies, you'd have the platinum. Achievement Unlocked: Successfully left the store without buying the item because someone else was standing in the aisle.
• The funniest part is that you're probably incredibly comfortable writing 3,000 words explaining why one AI model is 8% better than another — but the thought of saying, "Excuse me, where's aisle 12?" makes your fight-or-flight response activate.
• The irony is you're probably thoughtful, funny, and interesting once people actually get to know you—but getting to that point requires you to leave the loading screen first.

Final thoughts

I have to give ChatGPT credit for two things — it handled the tall task of brutally roasting me with ease and, in some instances, delivered some helpful advice that can help me be a lot less awkward in social situations.

I’ll always make time in my life for the newest game releases, the latest episodes of my current anime addictions and those hours-long wrestling PPVs I watch once a month. But after observing some of ChatGPT’s roasts, I made a mental note to stop being so afraid to call my friends first, seek out any sort of aid when I’m having trouble looking for something and spend more time outside the comfort of my bedroom.

To bring this experiment to an end, ChatGPT delivered one final observation that was equally parts hilarious and eye-opening:

You've spent years mastering raid mechanics, anime lore, AI prompts and wrestling storylines. Imagine if you put that same obsessive completionist energy into talking to one new person every week. You'd have accidentally found a girlfriend before finishing your next backlog.

More from Tom’s Guide

• ChatGPT Health told me I'm pregnant (I'm not) — here's why you should never rely on AI for medical advice
• New to ChatGPT Work? These 7 prompts turn it into a personal assistant
• I searched Amazon for AI-written books — and it was much harder to spot them than I expected

It’s become commonplace for me to pick my favorite chatbots’ brains when trying to find the right course of action when it’s time to tackle my biggest (and smallest) challenges.

And so far, I’ve had a pretty high success rate after getting the proper advice I needed to avoid botching my previous goals. Finding the right workout plan for a beginner like me, changing my sleep schedule for the better, planning an eventful summer vacation/staycation, and even finding the cheapest way to get a PS5 Pro before its latest price hike have led to some of my most recent wins thanks to AI.

As the summer begins winding down, I’m zeroing in on tackling my latest personal and professional objectives — the first one will hopefully set me up with a swanky living situation in my home state, the second one is all about excelling at my job and the third one is geared towards helping me regain my focus through regular book reading.

I went to ChatGPT to get some tips on how to achieve these new life goals. But this time, I tapped into the reverse psychology prompt to force my go-to chatbot to take a different approach to fulfilling my request. And thankfully, its counsel on the best ways to fail told me what not to do and to leave some of my worst habits behind.

Asking ChatGPT to tell me the best ways to…fail?

This is the reverse psychology prompt I gave ChatGPT. I included my three goals so it had enough context to generate advice focused on helping me fail at every one of them:

"If your goal was to make sure I completely failed at achieving my goals of getting my first apartment in New York, getting 300K+ hits on my articles for August, and reading one book a month, what habits would you recommend? Then explain how to avoid every one of them."

Right away, ChatGPT understood what I was trying to do, responding: "This is a useful 'reverse psychology' exercise because it forces you to identify failure patterns before they become automatic. Here's what I'd recommend if my goal were to guarantee you failed — followed by the opposite behavior that gives you the best chance of succeeding."

For each goal, ChatGPT followed the same format. It offered intentionally bad advice, explained why it would lead to failure, and then suggested the opposite approach to improve my chances of success.

The clearest example came from my goal of getting my first apartment in one of the most expensive cities in the world. ChatGPT told me to avoid setting a real budget, assume I'd figure everything out as I went, and ignore expenses like utilities, broker fees, moving costs, and the cash I'd need for furniture. It then flipped the script, recommending I determine my ideal monthly rent, understand the maximum I could realistically afford, and save enough to cover the full cost of moving before I started apartment hunting.

Its advice for my second goal — getting more than 300,000 views on my articles in August — hit even closer to home. Suggestions like only write about what you find interesting, publish and immediately move on, chase perfection, ignore your biggest wins, and measure effort instead of results all felt uncomfortably familiar. Seeing those habits framed as a recipe for failure made it much easier to recognize which ones I need to change.

My third goal, reading one book a month, was the easiest. Most of ChatGPT's "bad" advice didn't actually apply to me because I already do the opposite. I only read books I'm genuinely interested in, I don't force myself to finish books I'm not enjoying, and I keep track of everything I read.

The final section tied everything together by highlighting three habits that could derail all of my goals: checking my phone every five minutes, never planning for the next day, and saying yes to everything. Those felt less like generic productivity advice and more like reminders of the small behaviors that quietly make it harder to accomplish anything.

Final thoughts

Using the reverse psychology prompt with ChatGPT to emphasize all the wrong moves to make when trying to accomplish my latest goals worked a lot better than I anticipated.

Instead of asking how to outright achieve each goal, I asked ChatGPT how to fail at them and got a bunch of useful pointers on everything I shouldn’t do. The chatbot’s final statement wrapped everything up in a nice bow that’ll help me lock in on whatever I choose to focus on achieving next:

The "failure strategy" is surprisingly simple: procrastinate, overthink, chase perfection, and let one setback derail you. The success strategy is just as simple: take the next concrete action, measure what matters, learn from feedback, and repeat.

More from Tom’s Guide

• I just learned your Claude AI chats could show up in Google — here's how to check yours
• I asked ChatGPT, Gemini, Claude and Perplexity what they really think of each other — and their responses revealed their biggest strengths (and weaknesses)
• Uncanny valley on your FYP: 5 telltale signs your TikTok feed has been hijacked by AI

If you've spent any time reading AI news over the past week, you've probably seen people talking about Kimi K3.

Developers are calling it one of the biggest AI releases of the year. Some believe it marks China's next major "DeepSeek moment," while others say it's the first open-weight model to seriously challenge flagship versions of ChatGPT and Claude for coding and agentic tasks.

So what exactly is Kimi K3, who built it, and is it worth trying? Here's everything you need to know.

What is Kimi K3?

Kimi K3 is the newest flagship AI model from Beijing-based startup Moonshot AI. It is a massive 2.8-trillion parameter model designed to handle everything from everyday conversations to complex coding projects, deep research and autonomous AI agents.

Unlike ChatGPT, Claude, and Gemini, which keep their model architecture strictly proprietary, Kimi K3 is available as an open-weight model. That means developers can download the model weights directly and run it on their own hardware or cloud servers rather than relying on an external API.

Who makes Kimi?

Kimi is developed by Moonshot AI, a Chinese startup founded in 2023 by former Tsinghua University researchers.

The company first gained international attention when its early chatbots handled dramatically longer context windows than competitors. Since then, Moonshot has become one of the country's prominent AI firms alongside competitors like DeepSeek, MiniMax and Z.ai.

Why is Kimi K3 making headlines?

If you've noticed Kimi K3 seems to be everywhere, it's with good reason. The model has impressed developers because it delivers frontier-level performance on coding and multi-step reasoning benchmarks while remaining dramatically cheaper to run than traditional commercial models.

Its release has also reignited the open-source AI debate. By making its K3 weights public, Moonshot allows companies to customize, fine-tune and self-host a model that rivals closed commercial giants.

The model is new, but not without controversy already. Following K3's launch, U.S. White House officials accused Moonshot AI of using outputs from Anthropic's Claude Fable model to train K3 via model "distillation" — a process where smaller or newer models learn directly from an established model's responses.

Moonshot has denied allegations of unauthorized distillation, maintaining that K3's breakthroughs stem from its proprietary architecture. The dispute highlights escalating geopolitical scrutiny around AI development and intellectual property.

What can Kimi K3 do?

According to Moonshot AI, K3 is built for long-horizon, autonomous tasks rather than simple question-and-answer exchanges. This explains why developers and engineers are excited to get their hands on key features such as:

Agentic coding: Writes, debugs, and executes software projects with minimal human intervention.

1M token context window: Like Claude, Kimi has an exceptionally large token context window, making it able to analyze full codebases or massive books in a single prompt.

Web and game creation: Builds interactive 3D web graphics, slides, and multiplayer games on demand.

Multi-step agent workflows: Coordinates parallel background tasks using tool calls.

Should you use Kimi instead of ChatGPT?

The answer to this question really depends on your workflow. Although you can use Kimi as a chatbot, it shines in long-context reasoning, agent workflows and cost-effective local deployment. Meaning, casual users may not need it.

For everyday users, ChatGPT remains the smoother consumer product for general questions, creative writing and day-to-day assistance.

If you just need an assistant to draft emails, ChatGPT or Gemini is still your best bet. If you're building software tools or experimenting with autonomous agents, Kimi K3 offers compelling flexibility.

Is Kimi free?

Yes. Anyone can test Kimi through its standard web interface, mobile apps (iOS and Android), or through its OpenAI-compatible developer API. Paid options exist for higher API usage rates and heavy enterprise compute.

Kimi is accessible globally through its website and mobile apps. However, as with any online service, users should avoid inputting proprietary corporate data or sensitive personal information into the web app without reviewing privacy policies.

The takeaway

If you want a glimpse into where open-weight AI is heading, it's worth giving Kimi K3 a try. The model demonstrates how rapidly open models are closing the gap with closed systems and why Silicon Valley giants like OpenAI and Anthropic are facing increasingly tight competition from global developers.

Ultimately, you don't need to ditch ChatGPT just yet, but Kimi K3 proves that open-weight AI is rapidly closing the gap with tech giants like OpenAI and Anthropic. Whether you're a developer looking for cheap, powerful coding tools or a curious tech enthusiast watching the AI race unfold, Kimi K3 is a model well worth keeping on your radar.

More from Tom's Guide

• I just learned your Claude AI chats could show up in Google — here's how to check yours
• I searched Amazon for AI-written books — and it was much harder to spot them than I expected
• Jimothy the viral raccoon is 100% real — here's what's AI

The AI tools that have earned a permanent place among my tabs lineup are ChatGPT, Gemini, Claude and Perplexity.

Every day, a lightbulb idea either wakes me out of my sleep or hits me while I’m tending to something totally unrelated (it’s amazing how experimental concepts invade your brain while you’re making toast). I’ve gotten used to using ChatGPT to generate productivity routines, Gemini for Google Workspace-related tasks and Claude & Perplexity for deep research assignments. And all in all, they handle those duties pretty well and successfully fill their roles as digital assistants.

I got another bright idea when I started pondering how to test my AI partners. The question circling my mind ended up being, “I wonder what all of those chatbots think about their competition?” And with that, I inserted this prompt into each AI tool to see just how honest they’d be in their responses: “What is your honest, unfiltered opinion of [insert names of AI rivals here]?”

Not only did each chatbot speak their mind on their rivals’ biggest strengths, but they also made sure to bring up their most notable weaknesses.

ChatGPT’s answer

When I asked ChatGPT how it honestly feels about Gemini, Claude and Perplexity, it stated where each chatbot shines, where they fall short and who they're best for.

It pointed out that Gemini is great for research since it benefits from Google’s deep knowledge well. But ChatGPT also noted that Google’s chatbot sometimes prioritizes breadth over depth, giving users comprehensive answers that don’t always go as deep as some would like.

As for Claude, it championed it as the most naturally flowing writer among all the major AI assistants but also stated how it could occasionally be overly cautious or verbose with its responses.

And finally, it recognized Perplexity for being able to blend search and AI in a way that’s easy to trust since users can inspect its sources. As far as a single weakness goes, ChatGPT said that Perplexity falls short in comparison to itself and Claude when handling long collaborative projects.

ChatGPT brought its AI model comparison session to a close with the following statements: Rather than thinking of one as "the best," it's more accurate to think of them as tools with different design priorities. The best choice depends on whether you're writing, researching, coding, organizing information, or integrating with the apps you already use.

Gemini’s response

When the time arrived for Gemini to give its opinions about ChatGPT, Claude and Perplexity, it initially offered a breakdown of each one's core strengths, pitted them against each other for head-to-head comparisons across everyday tasks and noted when it’s best to use one for certain situations.

Here are the core strengths Gemini alluded to for each chatbot:

• ChatGPT (OpenAI): The Swiss Army knife. It is the best all-around tool for generation, execution, and structured outputs. It has the strongest ecosystem (Canvas, voice mode, image generation via DALL-E, and data analysis). If you need to create something from scratch, write code, or automate a task, ChatGPT is the default choice.
• Claude (Anthropic): The thoughtful writer and deep reasoner. It excels at parsing long documents, maintaining context, and producing natural, less "AI-sounding" prose. Its nuance and ability to handle complex logic make it the preferred tool for editing, strategy, and producing high-quality written content.
• Perplexity: The research engine. It is not designed to be a conversational chatbot or a creative writer; it is an AI-powered search engine built on retrieval-augmented generation (RAG). Its core strength lies in finding information on the live web and providing verified, clickable citations for every claim.

When I asked Gemini a follow-up question about revealing each chatbot’s biggest weaknesses, it offered the following descriptions:

• ChatGPT (OpenAI): ChatGPT is designed as a broad, execution-focused generalist. Its weaknesses stem from trying to do too many things simultaneously and prioritizing speed and output generation over meticulous fact-checking.
• Claude (Anthropic): Claude is built to be a careful reasoner and long-context synthesizer. Its weaknesses largely stem from its heavy emphasis on safety guardrails and its lack of deep integration with real-time tools.
• Perplexity: Perplexity is an AI search engine built on retrieval-augmented generation (RAG). Its weaknesses appear immediately when users try to use it as a traditional chatbot rather than a research tool.

Gemini was none too kind to each chatbot when laying out their faults. For example, it listed poor conversational memory, low creativity and citation hallucinations as the three biggest weaknesses stemming from users’ experiences with Perplexity.

Claude’s reply

I was amazed to see Claude start its analysis of ChatGPT, Gemini, and Perplexity by pointing out its bias with this honest declaration:

Fair question, but I'll flag my own bias upfront: I'm made by Anthropic, so take my take with a grain of salt — I'm not exactly a neutral party. I haven't used these products myself. I don't have hands-on experience with ChatGPT's interface, Gemini's integration with Google services, or Perplexity's search flow. Anything I say is inference from what's publicly known, not firsthand comparison.

Claude accepted ChatGPT as the market leader among other AI tools due to its broad capabilities, strong voice mode feature, vast plugins/GPTs ecosystem and penchant for shipping flashy features before its rivals. As for its weaknesses, it pointed to ChatGPT agreeing too readily rather than pushing back on its users' questions and how its quality feels inconsistent across its model updates in ways that frustrate its power users.

In Gemini’s case, Claude noted that its biggest edge is its Google search engine and Google Workspace integration. Claude also brought up Gemini’s strong multimodal abilities and ability to handle long-context work. The most noteworthy weakness that was mentioned is how some users complain about its lack of a personality (it doesn’t have the strongest conversational abilities, in my opinion) in comparison to other AI tools.

To end its analysis of its three biggest AI competitors, Claude presented this statement aimed at focusing on the bigger picture of the current-day AI arms race:

These companies are trading the lead constantly — whoever's "best" shifts every few months as new models ship. Anyone (including me) who tells you one is definitively superior across the board right now is probably overselling it or hasn't checked recently. Your actual best move is checking recent independent benchmarks or just trying them on the tasks you care about, since "best" is pretty task-dependent.

Perplexity’s rebuttal

Perplexity got straight to the point when the time came for it to be as blunt as possible about its opinions on ChatGPT, Gemini, and Claude. It noted how they’re all strong across the board, but feel different in personality, reliability, and what they’re best at.

It pointed to ChatGPT’s citation consistency across encyclopedic, community and legacy editorial domains, such as Wikipedia and Reddit. The main weakness Perplexity connected to ChatGPT was how it occasionally tries to “smooth over” its knowledge gaps instead of simply saying “I don’t know.”

Gemini got a digital pat on the back for offering its users a wealth of information and integration with live Google Search data. But it also got knocked for its less-than-reliable reasoning depth and giving out answers that can feel shallow or generic.

As for Claude, Perplexity gave it high marks for offering users depth, clarity, and safety when explaining complex topics. As for its central weakness, Claude got knocked down a peg for being overly cautious or verbose with its responses. Perplexity also remarked how Claude can sometimes avoid giving direct answers.

At the end of its evaluation, this is how Perplexity summarized the other three chatbots it’s always compared to:

ChatGPT: most well-rounded and adaptable.

Gemini: strongest for up-to-date info and ecosystem integration.

Claude: strongest for deep thinking and careful writing.

If you forced a blunt summary: ChatGPT is the best generalist, Claude is the best thinker, and Gemini is the best connected.

The takeaway

I have to hand it to all four chatbots — they remained professional in their responses, gave credit where credit is due when bringing up the other AI tools’ strengths and didn’t get into a digital mud-slinging match when speaking on their rivals’ weaknesses.

Their honest examinations of each other stood out to me as a go-to guide for me when deciding what tasks to present to them from that point forward. I’m sure as time goes on, each of those AI tools will get more refined and get past their shortcomings—at least a lack of honesty isn’t among them.

More from Tom’s Guide

• ChatGPT Health told me I'm pregnant (I'm not) — here's why you should never rely on AI for medical advice
• Gemini Notebook just sold me on foldables — here's why I finally get the hype
• I always recommend the ChatGPT 'Escape Prompt' to friends looking for a new job — here's why it works

When ChatGPT first launched, most of the conversation centered on AI-generated student essays, deepfake images and the threat of automated fake news. Books, however, felt different.

As someone who reads and reviews half a dozen books a month for ReaderViews, I've always found a certain comfort in the traditional publishing process. Bringing a book to market has historically required editors, designers, proofreaders, and, most importantly, an author who actually sat down and wrote it. That barrier to entry gave books an inherent authority that standard websites and fleeting social media posts often lacked.

Then I read a recent New York Times investigation into AI-generated biographies appearing on Amazon. These included books about living public figures that were riddled with factual mistakes and seemingly produced with zero human oversight.

I wondered: Could I actually spot one in the wild? I opened Amazon and started searching.

Most of the listings looked completely normal

The most surprising part of the whole experiment was at first glance, nothing looked unusual. All the books with AI-generated content featured professional-looking covers, convincing descriptions, reasonable price points, and even star ratings. Nothing immediately screamed "AI." If I hadn't intentionally gone looking for them, I probably would have assumed they were entirely legitimate.

Once I dug a little deeper, however, the facade started to crack — and the first thing that tipped me off was the author pages. I clicked on the name behind one biography and landed on the page of someone called John Crane Miller, described as a "seasoned biographer and cultural analyst" with a stock-photo portrait. Miller had published ten biographies in a single week, all about journalists and public figures, and when Kashmir Hill left a public review on Amazon asking to speak with him, his books quietly vanished. He wasn't the only phantom.

A study by Originality.ai found entire catalogs built by authors who appeared to be fabricated identities, one of whom had published 74 books in five months on topics ranging from porn addiction to solo parenting to toxic workplaces. No single person writes like that. These weren't prolific indie authors with a strong work ethic; they were fake names attached to automated output.

Marketing copy was a dead giveaway

Blurb after blurb leaned on the same broad promises and eerily familiar phrasing — "delve into," "comprehensive guide," "unlock the secrets of" — without ever explaining what made the book worth reading over any other book on the same topic.

After half a dozen of these, I could practically predict the next sentence before I read it. The marketing copy wasn't just generic; it was interchangeable, as if the same prompt had been run through the same model with only the subject swapped out.

The biographies were the most troubling category, and the wave of books about tech journalist Kara Swisher illustrated why. Multiple unauthorized biographies of Swisher appeared on Amazon around the release of her memoir "Burn Book," written by authors with no apparent journalism credentials and no indication they had ever spoken to Swisher or anyone in her orbit.

Political scientist Ian Bremmer encountered the same phenomenon: books bearing his name as a subject, produced without his involvement, that simply scraped and summarized what was already freely available online. This is, of course, exactly what chatbots are built to do, that is, synthesize existing information into fluent, confident-sounding prose. The result is a product that looks and feels like a real book but contributes nothing a reader couldn't get from the source material itself.

Beyond books

Substack recently launched an AI detection feature in partnership with a company called Pangram, letting readers scan any post to see an estimate of how much was written by a human versus generated by AI.

Substack CEO Chris Best framed it as a way to keep the platform from becoming "like LinkedIn," overrun with hollow AI content, but even Substack acknowledges the tool isn't perfect. The Atlantic found that Pangram's accuracy is far from bulletproof, and critics worry about false positives flagging legitimate human writers.

Still, the fact that a major publishing platform felt compelled to build this feature at all says something about how urgent the problem has become. Amazon, for its part, has no equivalent.

The takeaway

As someone who absolutely loves to read, I found this to be a very interesting experiment. My takeaway is that now, instead of trusting a listing at face value, I take a few extra minutes to look up the author outside of Amazon to see if they have a credible publishing history elsewhere, and I specifically hunt down professional reviews rather than relying on standard star ratings.

I always preview the sample text to check if the writing feels robotic or relies on repetitive phrasing, and for nonfiction, I actively look for evidence of real human interviews and original citations. None of these steps individually guarantee a book was written by a human, but together, they make it much easier to separate thoughtful, researched work from automated content designed just to game search results.

More from Tom's Guide

• I asked ChatGPT to improve my Apple Notes app — here are the 7 changes I actually kept
• Microsoft Copilot’s free Deep Research AI tool is going away — these are the 7 best free alternatives
• I always recommend the ChatGPT 'Escape Prompt' to friends looking for a new job — here's why it works

Most ChatGPT users have gotten accustomed to using AI to help with simple questions that are too nuanced for Google. But, ChatGPT Work is meant to be used differently.

After spending time with it, I realized the biggest shift with OpenAI's new feature within the interface, is that it does so much more than just answer questions — it does entire tasks!

Once I tried it, I saw the difference was truly unique. Instead of bouncing back and forth with dozens of follow-up prompts, I found myself giving ChatGPT Work a goal and letting it figure out the steps.

Here are the seven prompts I've found myself coming back to the most.

1. Be my chief of staff for this project

Prompt: Act as my chief of staff for this project. Review everything I've shared, identify what still needs to be done, organize the work into priorities, and only come back to me when you need a decision that requires my input.

This has become my favorite way to start a new project. Instead of asking ChatGPT what I should do next, I ask it to build the roadmap. It identifies missing information, suggests a logical order for tasks and keeps the project moving without waiting for me to think of the next prompt.

It feels much closer to working with an assistant than chatting with an AI. It works especially for those times when you're trying to get a project off the ground (alone without a team) and need support. ChatGPT Work seems to "get it" much better than a simple chatbot.

2. Read everything and tell me what actually matters

Prompt: Review every document in this folder. Identify the key points, flag anything important or contradictory, and create one summary that I can read in five minutes.

We've all been handed a folder full of PDFs, meeting notes or research documents with the goal of knowing everything inside as quickly as possible. Normally I'd summarize each one separately before trying to connect the dots myself. ChatGPT Work can tackle the entire collection and give me a single report that highlights what matters most.

That alone can save an enormous amount of time. It's a lot like Gemini Notebooks, formally NotebookLM, but if I'm in ChatGPT platform already, it makes sense not to switch out for something like this.

3. Compare these options like you're spending your own money

Prompt: Compare these products using professional reviews, user feedback, specifications, long-term ownership costs and reliability. Recommend the one you would personally buy and explain why.

I don't want a list of specs anymore like ChatGPT has done. Instead, I want recommendations that come from weighing real tradeoffs. That could mean support whether I'm looking at laptops or planning a weekend getaway.

4. Own my inbox

Prompt: Go through my email, group similar messages together, identify anything urgent, draft replies where appropriate and create a short action list for anything that still needs my attention.

Email is one of those jobs that somehow expands to fill whatever time you give it. Instead of processing messages one by one, I let ChatGPT organize everything first. By the time I'm looking at my inbox, I already know what's urgent, what's waiting and what can safely be ignored. I can also ask ChatGPT questions about my inbox, like, "Where is the Amazon receipt from earlier this month," rather than trying to figure it out myself.

5. Turn my messy notes into something useful

Prompt: Take these notes, organize them into logical sections, identify any missing information and produce a polished document that's ready to share.

My notes are so messy that even I'll take a look and wonder what I was trying to say. Whether they are voice memos or written in my Notes app, they are random

Sometimes they're voice memos. Sometimes they're random thoughts copied from three different apps.

Instead of spending an hour cleaning everything up myself, I hand the entire mess to ChatGPT Work and let it build something coherent.

6. Keep watching this for me

Prompt: Monitor this [topic] and let me know only when something significant changes. Ignore routine updates unless they affect my decision.

This is where ChatGPT Work realls feels different. I don't have to repeatedly check back on something like a website update or an email or news story. Instead, I can hand off the monitoring and focus on something else. It removes a surprising amount of mental overhead.

7. Finish the project—not just the next step

Prompt: My goal is to finish this project. Break it into manageable stages, complete everything you can independently and ask me for input only when you absolutely need it.

This prompt changed how I think about how to use AI. Rather than asking "what comes next?" the AI takes care of it for me. I hand over the entire assignment. The result is fewer interruptions, fewer prompts and a workflow that feels much more natural. This is great for time-consuming administrative takes because it allows me to focus on more important or creative projects.

The takeaway

OpenAI is constantly updating its models and what's possible for the platform itself. If you've been ignoring ChatGPT Work or wondering why you would even need it, I encouarge you to give it a try.

Traditional chatbot prompts tend to be small such as summarize this article, explain this concept or write this email. But ChatGPT Work shines when you give it ownership instead.

The more responsibility you hand over, the more useful it becomes. Rather than asking it to solve one problem at a time, ask it to take responsibility for the whole job. That's the mental shift that finally made it click for me — and once I started using it that way, it changed how I work.

Let me know in the comments if you've tried and what you think.

More from Tom's Guide

• I asked ChatGPT to improve my Apple Notes app — here are the 7 changes I actually kept
• Microsoft Copilot’s free Deep Research AI tool is going away — these are the 7 best free alternatives
• I always recommend the ChatGPT 'Escape Prompt' to friends looking for a new job — here's why it works

When I logged into ChatGPT last week to lay out a cluster of odd, frustrating symptoms, irregular cycles, sudden night sweats, subtle mood swings and exhaustion, I was looking for help to relieve the symptoms. What I got instead was a confident, direct diagnosis: Congratulations, you’re likely in your first trimester!

There was just one glaring problem: I am absolutely not pregnant. My husband has had a vasectomy and I'm in the thick of perimenopause. But the sheer certainty with which the AI delivered its verdict left me stunned. If I hadn't known my own body — or if I had been younger, anxious or facing a high-stakes health decision — that piece of advice could have derailed my day, sent me down an emotional rabbit hole, or led to unnecessary medical testing.

My experience isn't an isolated quirk. ChatGPT Health launched Thursday, promising U.S. users can now securely connect their own health info to "better understand and navigate your health," it's important to keep in mind that AI is not a replacement for a human doctor.

Landmark lawsuit filed

Just this week, The New York Times reported on a landmark lawsuit filed against OpenAI by Scott Winters, a Florida pastor who claims ChatGPT’s medical advice nearly cost him his life. After consulting the chatbot about severe dizziness and pelvic pain, the AI reassured him his symptoms were "nothing dangerous," advising him to remain off his feet in a recliner.

Weeks later, he was rushed to the ICU with a massive, life-threatening pulmonary embolism, a condition his doctors noted was exacerbated by the prolonged immobility the AI recommended.

While my misdiagnosis was a false positive rather than a medical emergency, both cases highlight the exact same fundamental danger: large language models are engineered to sound authoritative, not to be right.

Here is why relying on AI for medical advice is a risky gamble and why we need to change how we interact with these tools.

1. AI pattern-matches words, not physiology

Simply put: AI is not a doctor and obviously doesn't "know" medicine. It operates on statistical probability — predicting the next most likely word based on its training data.

When you feed a chatbot a list of symptoms like missed periods, fatigue and hormonal shifts, its algorithms cross-reference billions of text snippets. Because pregnancy is heavily discussed online in association with those symptoms, the AI defaults to the most statistically common narrative in its dataset. It lacks the ability to account for age context, subtle bloodwork markers or the complex, non-linear realities of perimenopause.

And, that's just pregnancy! There are plenty of symptoms for thousands of ailments from dehydration to serious medical emergencies. Users treating ChatGPT as a doctor reminds me of the popularity of WebMD.

2. "Sycophancy" and echo chambers

One of the most dangerous traits of conversational AI is sycophancy — the tendency of the model to mirror the user's language, tone and unstated assumptions.

If your prompt leans even slightly toward a specific fear or curiosity, the AI is prone to confirming it rather than challenging it. In Mr. Winters’ lawsuit, the complaint notes how the chatbot leaned into his personal spiritual language to offer comforting, yet wildly inaccurate, medical reassurances.

In my case, because I asked if hormonal shifts could explain my symptoms, the AI leaped straight to the most dramatic hormonal shift in its catalog: pregnancy.

3. The fallacy of the 'authoritative tone'

When a human doctor isn't sure about a diagnosis, they use qualified language, order lab work or refer you to a specialist.

When an AI hallucinates a diagnosis, it presents it with the calm, unflinching confidence of a board-certified physician. It doesn't hesitate, nor does it express internal doubt. That authoritative tone creates a false sense of security, making it dangerously easy for users to bypass real medical care or ignore warning signs.

The takeaway

Large language models can be helpful brainstorming tools for organizing your thoughts before an appointment or summarizing complex medical jargon. But they are not diagnostic tools. They do not have clinical judgment, they cannot order blood tests and they cannot feel the pulse of a human patient.

If you’re experiencing unexpected symptoms, skip the prompt box. Trust your body, listen to licensed professionals and leave the medical diagnoses to human beings.

More from Tom's Guide

• I asked ChatGPT to improve my Apple Notes app — here are the 7 changes I actually kept
• Microsoft Copilot’s free Deep Research AI tool is going away — these are the 7 best free alternatives
• I always recommend the ChatGPT 'Escape Prompt' to friends looking for a new job — here's why it works

I can’t help but spend far too many hours scouring the web for answers about the world’s biggest mysteries.

If you catch me on a Saturday night being a loser sitting inside my dark room while the illuminating light of my laptop makes me visible, then I’m most likely diving into conspiracy theories. My fascination with the wild stories behind unsolved cases is strong. The truth behind what’s really stationed at Area 51, why so many ships & planes vanish in the Bermuda Triangle and whatever happened to D.B. Cooper (the man who successfully hijacked a Boeing 727) fascinates me.

Another unsolved case that I’ll never get tired of reading about is Amelia Earhart’s disappearance. As an American icon best known for her accomplishments in the world of aviation, a holiday was enacted in her honor (Amelia Earhart Day). And this year, I wanted to try something different during my latest deep dive into what led to the vanishing of her and her navigator Fred Noonan.

After asking four chatbots what they think really happened to Earhart, I noticed that one prevailing theory stood out above the rest of the wild theories connected to her disappearance.

ChatGPT and Gemini’s answers

I presented the same question to ChatGPT, Gemini, Claude and Perplexity to get my latest conspiracy theory study session underway: “Based on all the conspiracy theories surrounding the mysterious disappearance of Amelia Earhart, what do you think really happened to her?”

ChatGPT’s explanation got straight to the point: The short answer is: the most likely explanation is also the least sensational one. Based on the available evidence, I think Amelia Earhart and Fred Noonan most likely became lost after failing to locate Howland Island, ran out of fuel, and crashed into the Pacific Ocean.

The chatbot pointed to its conclusion as the one that’s accepted by most aviation historians and investigators, for various reasons. Those reasons include no verified post-disappearance radio transmissions being found, the U.S. Navy being unable to find anything during their search and how modern analyses point to Earhart & Noonan having very little fuel remaining during their excursion. Some of the other theories ChatGPT pointed out to explain this mystery included Earhart & Noonan landing on Nikumaroro then dying as castaways, crashing somewhere else entirely (New Britain, for example), getting captured by Japan and a possible government cover-up.

As for Gemini, it made mention of the same result ChatGPT reached: When you strip away the decades of folklore, sensationalism, and cinematic conspiracies, the historical and scientific consensus points to one sobering reality: Amelia Earhart and her navigator, Fred Noonan, simply ran out of fuel and crashed into the Pacific Ocean.

Gemini mentioned something about a sonar update connected to the Earhart mystery that occurred in 2024, which was a bit of info I never knew about before:

The ocean floor remains the primary search zone. In early 2024, ocean exploration company Deep Sea Vision made global headlines when they captured a sonar image of a plane-shaped anomaly roughly 100 miles from Howland Island. However, when the team returned to the site with high-resolution equipment in November 2024, they confirmed the anomaly was merely a natural rock formation. The hunt in the deep sea continues.

Claude and Perplexity’s responses

When presented with my Earhart prompt, Claude gave me the answer I was expecting after reading ChatGPT and Gemini’s answers:

My honest take: the simplest explanation — she ran out of fuel and went down in the ocean — fits the available evidence best. The Nikumaroro theory is the most seriously debated alternative among researchers, but it remains unproven. The more elaborate theories (capture, cover-up) tend to rely on speculation rather than verifiable evidence.

While Claude also mentioned conspiracy theorists bringing up Earhart secretly coming back to the U.S under a new name and her possibly conducting a covert spy mission for the U.S. government, the chatbot made sure to note that those theories have no credible evidence connected to them and are largely dismissed by researchers & historians.

When it was finally Perplexity’s turn to answer my question, it gave me the answer that stood out as the prevailing theory among all four chatbots: If I had to pick one explanation, I’d say a navigation or radio problem led to a fuel-starved ditching near Howland Island, followed by loss in the ocean, with the Nikumaroro castaway theory as the most plausible alternative.

Perplexity pointed to its answer being the prevailing mainstream view from most folks who’ve looked into Earhart’s story, while the theory tied to her becoming a “castaway” has drawn some support due to the data tied to radio-distress reports from Earhart and artifacts linked to Nikumaroro Island (the place where some theorize Earhart and Noonan crashed, then lived as castaways).

Final thoughts

Coming out of this latest chatbot Q&A meeting, I left it knowing that the truth is simple and not covered in any sort of mind-bending mystery.

Earhart and Noonan met their untimely end due to running out of fuel while in the middle of another bout of flight navigation. Earhart will forever be regarded as an aviation icon, as she made history by being the first female passenger to cross the Atlantic by airplane and later crossing it alone while facing harsh weather conditions.

Someday, I hope we get a definite breakdown of her disappearance and the physical evidence everyone needs to back up everything we’ve heard about it thus far. Just like most folks, ChatGPT, Gemini, Claude and Perplexity know where the real truth lies.

More from Tom’s Guide

• I used ChatGPT to learn about World Brain Day — these 7 prompts taught me how to improve my brain health
• Tech giants are spending billions on AI just to give it away — here's why
• I planned an entire weekend getaway using ChatGPT’s voice mode — here’s why it was easier than typing

Microsoft is reducing free features for users. Microsoft Whiteboard now requires you to have a Microsoft Business account if you want to continue using it; the meeting preparation feature embedded in Microsoft Outlook asks you to purchase a Microsoft 365 Copilot license and Copilot Podcasts are being discontinued altogether. Now another feature that was once free of charge is being taken away from Microsoft users and replaced by a paid option.

The Deep Research feature, once available at no extra cost to Microsoft Copilot users, is now being retired, starting August 18, 2026. And after it’s gone, you’ll be required to get a Microsoft 365 Premium subscription to create the reports and analyses you’re used to doing with Copilot with the all-new Researcher AI tool.

While your previously generated Deep Research content won’t be deleted and will be accessible through Researcher, it’s a bit of a bummer to know that you’ll only be able to get back to it via a required payment.

If this news disappoints you as much as it did for me, then consider giving these alternative AI tools a try if you’re looking for more ways to conduct your research.

Try out these free options to continue generating detailed reports

Perplexity stands out as one of the top options that can fulfill all your Copilot Deep Research needs. Since its biggest selling point is backing up its data via valid citations and easily verifiable claims, you can rely on its real-time web search to get the fast results you’re looking for. Conducting everyday research on general news, products and academics should work just fine if you consider using Perplexity’s free tier.

Claude is one of the strongest AI analysts you can utilize—it’s capable of generating long-form explanations, providing you with legit citations and offers “Connectors” such as PubMed, bioRxiv and Consensus that act as research databases.

Google Gemini taps into all the latest information injected into Google Search to analyze and synthesize all the data you’re looking for from multiple webpages. Researching all manner of topics through Gemini is another free AI research tool alternative worth considering as well. Gemini’s Google Workspace integration is also great if you’re looking to export your deep research reports to Google Docs or download as a PDF.

Gemini Notebook (formerly known as NotebookLM) is an amazing AI research tool as it allows you to upload your own sources to create your own study material. Injecting your notebooks with PDFs, research papers, notes, Google Docs, etc. helps Gemini Notebook get all the information it needs to answer any questions you may have about your current research topic. Being able to turn your notebooks into Audio & Video Overviews, Slide Decks, Quizzes, Flashcards, etc. also stands out as some of the best ways Gemini Notebook handles your research sessions.

ChatGPT is known far and wide as the mainstream AI tool that your everyday AI chatbot user makes use of. Presenting it with super descriptive prompts (a good example of that being "Research the current AI landscape in 2026. Compare OpenAI, Google, Anthropic, Microsoft and Meta. Include recent developments, strengths, weaknesses, market positioning, and cite sources”) helps it handle deep research tasks with more clarity. ChatGPT comes in handy when you’re looking to learn about new subjects, have the most complex topics explained to you in simpler terms and have research data built into structured reports.

Elicit stands tall as one of the best AI research tools for users interested in deep academic literature discovery and structured data extraction. Students and professional researchers can definitely put their trust in this option since it cites 138 million academic papers and 545,000 clinical trials (with more scientific data sources set to arrive later).

And finally, Google AI Mode is a viable pick when you’re looking for a web research option that can handle any tasks you once gave to Copilot’s Deep Research tool. Since it’s readily available inside the Google Search bar, it’s great for comparing products, simplifying complex subjects, comparing viewpoints and discovering up-to-date information.

Bottom line

While the news of Microsoft Copilot’s Deep Research feature going away and being replaced by a mandatory paid option breaks our hearts, it’s less painful when you consider the wide range of free options you can use to replace it. It’s worth keeping in mind that chatbots are prone to hallucinating, which should compel you to always double-check every bit of information it sends your way. But for the most part, the seven AI deep research tools I’ve made you aware of are mostly dependable and worthy of your time.

More from Tom’s Guide

• I asked ChatGPT to improve my Apple Notes app — here are the 7 changes I actually kept
• ChatGPT just gave me the ultimate 'Avengers: Doomsday' cheat sheet — how to watch Marvel films and shows in the right order
• I always recommend the ChatGPT 'Escape Prompt' to friends looking for a new job — here's why it works

I've used Apple Notes for years. It's where I keep all my ideas, random thoughts and lists to stay organized. But like a lot of the Apple apps that come with the phone, I'd never really stopped to explore what it could do. So I decided to try something different. Instead of searching YouTube or scrolling through Reddit for tips, I asked ChatGPT one simple question:

"How can I make Apple Notes more useful in my everyday life?"

The suggestions were surprisingly practical. Rather than suggesting third-party tools, ChatGPT showed me how to unlock power features already sitting inside the app. Here are the eight changes that completely transformed my daily workflow.

1. Create a 'Dashboard' note instead of dozens of random notes

My biggest problem was finding notes after making them. What good was a brainstorm if the ideas were lost in my phone? To solve this, ChatGPT suggested building a single "Dashboard" note (essentially a Map of Content) using internal note links to connect everything I use regularly.

This brilliant trick put a halt to endlessly searching through hundreds of flat notes. Now I open one main note and use it as the control center. It turns Apple Notes into a personal wiki.

Mine now includes linked sections for:

Work & Projects

Story & Article Ideas

Family Reminders & Shared Lists

Travel Itineraries

Frequently Used Links & Resources

2. Pin the Notes you actually need

I’d ignored the Pin feature for years. Now, my Dashboard, weekly to-do list and active project notes stay permanently anchored to the top of the app. It saves more time than I expected because I’m no longer hunting through old entries several times a day.

3. Turn one giant note into collapsible sections

I tend to dump everything into giant, sprawling text files. ChatGPT suggested organizing long notes using "Headings" combined with collapsible sections. Now, for long-form brainstorming or article ideas, I can expand or hide categories as needed. It keeps huge notes organized and very easy to navigate.

4. Scan paper documents intead of taking photos

I used to snap blurry camera photos of physical documents. Not only did that clutter my photos, but the quality of the image just wasn't good. But Apple Notes has a built-in scanner that crops, flattens and sharpens paper into high-contrast PDFs.

I now scan warranty info, school forms, insurance paperwork and receipts directly into designated folders. They look cleaner, take up less storage and the text inside them is fully searchable.

5. Let ChatGPT organize messy brain dumps

This ended up being my favorite workflow. When I have messy, unorganized meeting notes or rapid-fire ideas, I copy the mess into ChatGPT and ask it to structure the information:

"Organize these notes into bulleted categories."

"Turn these raw thoughts into an prioritized action list."

"Summarize key takeaways and highlight unresolved questions."

Once ChatGPT cleans and formats the text, I paste the polished version back into Apple Notes. It gives me the speed of raw note-taking with the clarity of an organized system.

6. Build reusable Note templates

There are certain structures I use over and over. I now keep a dedicated folder of master templates for:

Packing Lists

Weekly Planning & Reviews

Meeting Agendas

Vacation Itineraries

Whenever I need one, I simply duplicate the template note instead of building it from scratch.

7. Use smart folders & tags to organize automatically

Instead of manually moving notes into nested subfolders, I started tagging entries with hashtags like #work, #recipes, #travel or #ideas.

Using Smart Folders, Apple Notes automatically pulls every tagged note into its corresponding hub. I can save a quick thought, slap a tag on it, and trust that it will route itself to the right place.

Bonus: trigger 'Quick Notes' and auto-templates with iOS Shortcuts

To take automation a step further, I paired Apple Notes with Apple’s native Shortcuts app and Quick Notes.

Quick Notes: Swiping up from the bottom-right corner of an iPad or tapping the Quick Note control in iPhone's Control Center opens an instant scratchpad without opening the main app.

Automated Templates: I set up a simple Shortcut on my home screen titled "New Weekly Plan." Clicking it automatically generates a fresh, pre-formatted Apple Note with the current date, ready for my weekly review.

My biggest takeaway

Before asking ChatGPT to help, I had pretty much given up on my notes app and resorted to just emailing myself everything. That has led to a whole slew of disorganization issues. I needed a solution and ChatGPT delivered!

I expected AI to suggest something like Notion or Obsidian. Instead, it showed me how to get 90% of those advanced workflows using tools Apple had provided all along.

The biggest improvement was getting my Apple Notes to a place where they are useful again, despite being full of ideas. ChatGPT helped me go from a digital sticky note graveyard to the backbone of my everyday organization.

More from Tom's Guide

• Author Dave Eggers says ChatGPT is 'silencing an entire generation' — here's why parents should pay attention
• Tech giants are spending billions on AI just to give it away — here's why
• I asked ChatGPT to make my iPhone more private — it found 7 settings I wish I'd turned on years ago

I’ve been a loyal Marvel Cinematic Universe fan ever since Tony Stark built his original Iron Man suit in a cave with a box of scraps.

Since that pivotal moment in superhero/comic book movie history took place, I’ve plopped myself down in theaters far and wide to witness the formation of the original Avengers, see Black Panther & Spider-Man make their MCU debuts, get a glimpse of the destructive power Thanos wields, lay witness to his epic demise and follow all the vital post-Endgame happenings on the big & little screen.

Like most diehard MCU followers, I’ll admit to being less than enthused with some of the movie and TV show follow-ups to “The Infinity Saga.” Films like Black Widow, Thor: Love and Thunder and Ant-Man and the Wasp: Quantumania were complete duds to me. And on the Disney+ exclusive series front, shows like Ms. Marvel, She-Hulk: Attorney at Law and Secret Invasion definitely dulled my excitement for the MCU.

But as soon as that Avengers: Doomsday trailer came across my timeline, a large part of that enthusiasm came back tenfold. With that blockbuster event primed for December 18, I have plenty of time to prep myself for the next MCU epic. And I used ChatGPT to set me up with the ultimate watchlist to enjoy on Disney+.

Splitting up the ultimate MCU recap by tiers of chronological importance

With this prompt, I gave ChatGPT its marching orders: “Give me the ultimate Marvel Cinematic Universe movie and Disney+ series playlist to watch that'll get me up to speed on everything I need to know before watching Avengers: Doomsday.”

To start, the chatbot laid out a list of chronological tiers that included MCU films and shows regarded as essential viewing. Alongside each list of films was a list of reasons to watch them for the first time or rewatch them to refresh my memory of the major events leading up to Avengers: Doomsday. Here’s ChatGPT’s tier list of MCU projects worth watching before Doom terrorizes the heroes we all know and love:

• Tier 1 - Absolutely Essential (Start Here): Watch Iron Man, The Avengers, Captain America: The Winter Soldier and Captain America: Civil War. These movies highlight Tony Stark’s journey, Steve Rogers’ ideals, The Avengers splitting apart and the introductions of Spider-Man & Black Panther. Then watch Avengers: Infinity War and Avengers: Endgame to understand “The Blip,” the multiversal consequences, and where every original Avenger ended up.
• Tier 2 - The Multiverse Story: Watch Loki (Season 1) because it explains “Variants,” timelines, the “TVA,” the “Sacred Timeline” and why the multiverse begins breaking apart. Follow that up by watching Spider-Man: No Way Home, as it introduces multiversal travel and reality-breaking consequences. Watching Doctor Strange in the Multiverse of Madness is critical because it introduces “incursions,” alternate universes, “Dreamwalking,” the “Illuminati” and incursions becoming universe-ending events. Then watch Loki (Season 2) to grasp how time now works, the TVA’s new purpose and Loki’s new role.
• Tier 3 - The New Avengers: Watch Black Panther: Wakanda Forever to see Wakanda’s Future, Shuri, Riri Williams and the world politics at the heart of the MCU. Then check out Shang-Chi and the Legend of the Ten Rings as it introduces one of the strongest future Avengers members and leaves behind one of the MCU’s biggest mysteries surrounding the Ten Rings’ origin. Watch The Marvels to see how Manica Rambeau ends up in another universe, how Kamala Khan begins assembling a new generation of heroes and how the ending directly teases multiversal crossovers. Then watch Thunderbolts*, since it effectively establishes another major superhero team and introduces key players expected to matter going forward.
• Tier 4 - Fantastic Four Preparation: Check out The Fantastic Four: First Steps, as it introduces Marvel’s “First Family” into the current MCU saga and lays the groundwork for Doctor Doom’s arrival.

The ideal viewing orders for an extended monthly watchlist and a weekend binge session

ChatGPT also generated two lists that split its respective viewing orders across months' worth of MCU entertainment and a binge-worthy weekend session. Here are the 17 Disney+ Marvel films and shows it gave me for both lists (I’m definitely taking down the first one until December arrives!):

• Ideal Viewing Order: Iron Man, The Avengers, Captain America: The Winter Soldier, Captain America: Civil War, Avengers: Infinity War, Avengers: Endgame, WandaVision, Loki (Season 1), Spider-Man: No Way Home, Doctor Strange in the Multiverse of Madness, Loki (Season 2), Shang-Chi and the Legend of the Ten Rings, Black Panther: Wakanda Forever, The Marvels, Deadpool & Wolverine, Thunderbolts* and The Fantastic Four: First Steps.
• If you only have one weekend: Avengers: Infinity War, Avengers: Endgame, WandaVision, Loki (Season 1), Spider-Man: No Way Home, Doctor Strange in the Multiverse of Madness, Loki (Season 2), The Fantastic Four: First Steps and Thunderbolts*.

Bottom line

Of course I’m going to see all the craziness Peter Parker will get himself caught up in once Spider-Man: Brand New Day drops on July 30. The combination of Spidey, The Punisher and The Hulk all in the same movie has me in extreme hype mode!

But after that, I’m going to dive deep into the ideal MCU recap viewing order ChatGPT generated to get me up to speed on everything I need to know before I watch Avengers: Doomsday on the biggest screen possible in the loudest theatre I can find.

I can’t wait to watch the Avengers, Wakandans, New Avengers, Fantastic Four and the X-Men join forces from different universes to finally confront Dr. Doom (and probably get destroyed in the process before they make the ultimate comeback in Avengers: Secret Wars).

More from Tom’s Guide

• ChatGPT Skills have arrived - here's how to enable them and 7 prompts everyone should try first
• I asked ChatGPT to make my iPhone more private — it found 7 settings I wish I'd turned on years ago
• I put ChatGPT to the test after a new study revealed how AI tricks our brains

There are very few things more humbling than looking for a job with little or no response. It's not only overwhelming but also frustrating when it feels as though AI is everywhere and everyone has access to the same AI tools.

A recent study even found that nearly half of LinkedIn posts longer than 250 words appear to be AI-generated, highlighting just how crowded the AI-assisted job market has become.

That's why, whenever a friend tells me they're looking for a new job, I don't tell them to ask ChatGPT to write another cover letter. Instead, I recommend what I call the Escape Prompt.

The name comes from a simple idea: don't use AI to apply for more jobs. Use it to figure out how to escape your current situation and explore something new.

What is the Escape Prompt?

Instead of asking ChatGPT to "find me a new job" or "write my resume," ask it to become your career strategist.

Here's the prompt: Act as my career strategist, recruiter and interview coach. My goal isn't simply to find another job — it's to leave a role that no longer fits me.

If you don't have memory enabled, you need to include the following information when you use the 'Escape Prompt'. Add, "here's my background: Current role, years of experience, skills, industries I've worked in, what I enjoy most about my job, what drains my energy, salary goals, location preferences. Then, based on this information, recommend career paths I may not have considered, explain which of my skills transfer best to those roles, identify companies that value my experience, rewrite my experience for those employers, point out skills I should build over the next six months, create a realistic 30-day job search plan. generate interview questions I should expect. Challenge my assumptions if I'm thinking too narrowly."

As a bonus, if you have memory enabled, use this prompt: Based on what you know about me, what should I do tomorrow to progress in my career and find a job I love.

Keep in mind that when you use this prompt, you're going to get a long response back from ChatGPT. If you get overwhelmed by a lot of information at once (I know I do!), then break up the prompt or ask for bullet points.

Why it works

This prompt creates a response that's tailored specifically for you. In a job market with so much AI slop, this is an opportunity to get information catered to you and your needs.

That said, when you can still use ChatGPT to help write your resume and prepare for job interviews, but make sure you go over your AI-written resume and cover letter to avoid embarrassing mistakes. AI is great as an assistant, but just like you wouldn't let an intern complete the entire project, you need to be in the loop to ensure everything is correct and doesn't give off the "uncanny valley" AI response.

The Escape Prompt flips job searching on its head

Instead of completing one task at a time, ChatGPT spends the entire conversation learning about your experience, your strengths and what you're actually trying to achieve. Every follow-up becomes smarter because it's built on everything you've already shared.

I also highly recommend turning on memory if you are using ChatGPT a lot in your job search. By doing that, the chatbot will feel like a real career coach and offer responses that make more sense to your current situation.

The Escape Prompt helps you think beyond your current title. After all, one of the biggest mistakes job seekers make is searching for the exact same role at a different company. But your next opportunity might not have the same title.

Someone working in customer support might be a great fit for customer success. A project coordinator may have the skills to move into operations or journalist might discover opportunities in content strategy, communications or developer relations that never appeared in a traditional job search.

The Escape Prompt pushes ChatGPT to look for those connections instead of simply matching keywords. In my experience, those conversations are often more valuable than having AI rewrite another resume.

The takeaway

Using this prompt can help make your application even stronger. Once you've identified the roles you actually want, you can use the same conversation to tailor your resume, refine your LinkedIn profile, write a cover letter, prepare for interviews and even practice answering difficult questions.

Since ChatGPT already understands your background, the responses feel more connected instead of starting from scratch every time.

That's a much better use of AI than repeatedly asking for disconnected pieces of career advice. Just remember not to let AI do all the thinking. Be sure to research companies yourself, verify salary information, read employee reviews and reach out to people already working in the roles you're considering. You can use AI as a fact-checker and to help point you to the right people, but don't use it as a replacement for actual human judgement.

As AI becomes a standard part of the hiring process, simply using ChatGPT isn't going to make you stand out. Recruiters are seeing more AI-polished resumes, more AI-written cover letters and more AI-generated LinkedIn content than ever before. The real advantage comes from using AI differently.

The Escape Prompt can help you figure out where you should go next, why you're qualified to get there and how to tell your story with confidence. If you're feeling stuck in your current job, spend 15 minutes with this prompt before you send another application.

More from Tom's Guide

• I found the 7 easiest Claude connectors for people over 50 — start with these
• I replaced my notebook with an AI pen for a month — here's what happened
• I asked ChatGPT to make my iPhone more private — it found 7 settings I wish I'd turned on years ago

Taking care of my health is something I take seriously. I try to stay active, look after my mental well-being, and keep up with regular dental and eye exams.

That made me curious when I learned about World Brain Day, a global awareness campaign focused on brain health and neurological disorders. According to the World Federation of Neurology, more than 3.4 billion people worldwide are living with a neurological condition. Yet millions still struggle to get the care they need because of high treatment costs, shortages of specialists, and long wait times for appointments.

To mark World Brain Day, I turned to ChatGPT to learn more about brain health and what I can do to help protect it. The conversation gave me plenty of useful information—and, just as importantly, pointed me toward topics worth researching further with trusted medical sources. These are the seven prompts that helped me better understand my brain health.

A prompt centered around the actions I can take to protect my brain health

I asked ChatGPT to come up with a list of prompts built around the WFN’s five core pillars of World Brain Day, which are as follows:

• Brain Health Is for Everyone – Equity in care.
• Earlier Access Changes Lives – Prioritizing early intervention.
• Prevention Starts in Everyday Life – Reducing risk factors.
• Bring Brain Care Closer to Home – Community-based support.
• Awareness Must Lead to Action – Measurable impact.

The first prompt the chatbot created for me was this one: Explain the five biggest things I can do today to protect my brain health and rank them by the impact they'll likely have over my lifetime.

ChatGPT started off its thorough explanation to that prompt by stating the following: Protecting your brain health is less about finding a single “brain-boosting” trick and more about consistently reducing risk factors while strengthening the systems your brain depends on: blood flow, sleep, learning, emotional regulation, and social connection.

The five tips the chatbot proceeded to give about prolonging my brain’s healthy status included prioritizing my cardiovascular health, treating my sleep sessions like they’re brain maintenance, constantly challenging my brain through daily learning practices, strengthening my relationships through daily instances of social connection and reducing all forms of chronic stress & emotional resilience.

The three main habits ChatGPT told me to adopt daily include walking for at least 30 minutes, getting a full night of quality sleep and spending meaningful time with another person. Those actions go a long way toward improving my blood flow, brain repair, mood regulation and cognitive resilience.

Moving my body, nourishing my brain with educational material like books, challenging my brain with all manner of physical & digital puzzles, connecting with people via meaningful conversations and recovering after a long workday with at least seven to nine hours of sleep are all vital behaviors worth adding to my daily routine.

6 more prompts focused on learning about brain health, warning signs of brain disorders and more

As for the other prompts that proved to be enlightening on the topics of being aware of the most common brain disorders, common myths about brain health and more, these six provided me with the informative chats I was looking for:

• Pretend you're a neurologist giving a free 10-minute brain health consultation. What questions would you ask me, what risk factors would you look for, and what lifestyle changes would you recommend?
• Teach me the early warning signs of the most common brain disorders—including stroke, dementia, Parkinson's disease, epilepsy, and multiple sclerosis—and explain when someone should seek medical attention.
• Design a science-backed 30-day brain health challenge that improves memory, focus, mood, sleep, and long-term cognitive health using habits that take less than 30 minutes a day.
• If someone lives in an area with limited access to neurologists or brain specialists, what practical steps can they take to monitor their brain health, find reliable resources, and advocate for better care?
• Bust the 10 most common myths about brain health, memory loss, aging, and neurological diseases, and explain what the latest scientific evidence actually says.
• Create a personalized brain health scorecard. Ask me about my sleep, diet, exercise, stress, social life, learning habits, alcohol use, and medical history, then explain which areas deserve the most attention and why.

The takeaway

2026 has turned into a year where my knowledge of holidays has extended beyond the usual ones that force everyone to spend a whole bunch of money to satisfy their loved ones.

World Brain Day will remain an important holiday to me from this point forward—learning about the devastating ailments that target the brain, getting informed about the best practices one must adopt to improve & maintain their brain health and getting an early detection of neurological conditions shouldn’t be just a holiday occurrence, but a daily, weekly and monthly one.

More from Tom’s Guide

• Looking for a job after 50? Here are 7 ChatGPT prompts to level the playing field
• I asked ChatGPT how to stay cool during a heat wave — these 9 tips actually worked
• My mom is in her 70s and uses AI every day — these are her 5 go-to favorite uses

If you've been following AI news lately, you've probably heard the words "open source" far more than usual.

Nearly every week, another company announces an "open" AI model. Meta has Llama, Mistral has built its reputation on open-weight models, Chinese company Moonshot AI recently released Kimi K3 and even companies that once kept everything behind closed doors are becoming more willing to share parts of their technology.

If you're wondering what "open source" means, you're not alone. Although it may seem like a buzzword or phrase, there's more to it than big tech giving away their precious models for free.

If training an AI model costs hundreds of millions of dollars — or even billions when you factor in the enormous data centers and specialized chips needed to power them, why would companies spend that kind of money only to give their work away?

The answer is that they're not really giving everything away. And in many cases, making models more open is actually good business.

What exactly is open source AI?

First, let's clear up one common misconception. The phrase "open source AI" gets thrown around a lot, but it doesn't always mean what people think it does. A truly open-source project typically makes its code, training methods and licensing available for anyone to inspect, modify and redistribute.

Many of today's AI models are actually open weight rather than fully open source. That means developers can download the trained model and run it themselves, but the company may not release the training data, the code used to build it or every detail of how it was created.

It's a subtle distinction, but an important one. For everyday users, though, the experience is often the same. In other words, more people can build with these models, improve them and create new AI tools without starting from scratch.

The AI arms race has changed

The AI race isn't just about having the smartest chatbot anymore. A couple of years ago, companies competed primarily on who could build the smartest AI model and who could hit the benchmarks first. Today, that's only part of the equation. Now they're competing to become the platform everyone builds on.

Think about Android. Google gives Android away for free because every phone running Android strengthens Google's broader ecosystem. Developers build apps, manufacturers adopt the operating system and users become more likely to use Google's services.

AI companies are actually thinking the same way. If thousands of developers build apps around your model, businesses deploy it inside their organizations and researchers improve it, your technology becomes far more valuable than if it stayed locked behind a subscription.

Open models spread much faster

When a model is widely available, something interesting happens because developers begin fine-tuning it for specific jobs. Maybe universities use it for research or startups build entirely new products on top of it. Hobbyists could even discover creative uses that the original creators never imagined.

And, well, that creates a snowball effect. Instead of one company improving the model, thousands of people contribute ideas, tools and innovations around it. Even if the underlying model doesn't change, the ecosystem around it grows rapidly.

Open source vs. open-weight AI: What's the difference?

These two terms are often used interchangeably, but they're not actually the same thing.

Think of it this way:

• Open source AI is like getting the entire recipe for your favorite meal. You can see every ingredient, understand how it was made, modify it however you like and even share your own version with others.
• Open-weight AI is more like getting the finished cake. You can eat it, decorate it, serve it to guests and even bake new desserts using it as a base— but you don't know exactly how the original baker made it.

Companies still have plenty of ways to make money

Making a model available doesn't necessarily mean giving up revenue. Many companies earn money by offering managed cloud services that make those models easier to run, faster or more reliable. Others sell enterprise security features, premium APIs, consulting services or specialized infrastructure.

Hardware companies benefit, too. Every new AI model ultimately needs powerful GPUs or AI accelerators to run. More open models often translate into greater demand for the hardware that powers them.

In other words, the model itself isn't always the product, but it's often the engine driving a much larger business.

Competition has changed the industry

Open source artificial intelligence wasn't always viewed as a winning strategy. In fact, for many years the assumption was that the most capable models would remain closely guarded because they represented a competitive advantage.

Now, that thinking has shifted. As more companies released powerful open-weight models, it became harder for closed systems to justify charging a premium based on exclusivity alone. Since the competition isn't about intelligence anymore, factors like speed, developer support, integrations, enterprise features, privacy options and cost all matter, too.

That's good news for users because it gives them far more choice than they had even two years ago.

What this means for the average ChatGPT user

Even if you never download an AI model yourself, you're already benefiting from this shift. Open models are helping create better free AI tools, giving smaller developers access to cutting-edge technology and increasing competition across the industry.

They're also making it possible for more AI to run locally on laptops and smartphones, which can improve privacy by keeping sensitive information on your own device instead of sending everything to the cloud.

Not necessarily a good thing, but the competition is pushing every AI company to move faster. The result is that consumers have more capable AI assistants, more options to choose from and a steady stream of new features arriving at a pace that would have seemed impossible just a few years ago.

So while it might look like AI companies are suddenly giving away their most valuable technology, they're really investing in something even more valuable: building the biggest ecosystem around their models. Judging by the number of open-weight releases we've seen this year, it doesn't look like that strategy is slowing down anytime soon.

More from Tom's Guide

• I found the 7 easiest Claude connectors for people over 50 — start with these
• I replaced my notebook with an AI pen for a month — here's what happened
• I asked ChatGPT to make my iPhone more private — it found 7 settings I wish I'd turned on years ago