Club Benefits
The speed at which ChatGPT answers even my most layered questions is genuinely impressive.
Whether I’m asking for the best-reviewed video games of the year, which states have the top theme parks, or how a new MacBook stacks up against competitors, it delivers detailed responses in seconds — often enough to help me make a faster, more informed decision alongside my own research.
But speed isn’t the same as accuracy. As useful as ChatGPT is in both my personal and professional workflow, I never take its answers at face value. Like most chatbots, it can be confident, polished — and occasionally wrong.
Article continues belowAI moves fast, sounds convincing and presents information clearly. But if you rely on it without verifying what it says, it’s easy to miss subtle errors or outdated details.
That’s why I’ve built a simple system I use every time I work with ChatGPT. It lets me keep the speed — without sacrificing accuracy.
Keep an eye out for ‘high-risk points’ and use two sources for backup
As soon as ChatGPT lays out its detailed answers to my latest inquiry, I quickly look through them all to pick up on details that I mark as “high-risk points.” I define those as pieces of information that are presented as factual, but may be incorrect and lead to editorial errors if I rely upon them as usable data.
Those high-risk points include:
- Numeric details such as statistics, percentages, and prices
- Specific dates and timelines
- Quotes or attributed statements
- Proper names for companies, products, people, and groups
When something feels even slightly off, I assume it is — and move quickly to verify it.
My rule is to check two additional sources.
First, I look for an official source — a company website, press release or documentation that directly supports the claim. Then I cross-check it against a reputable publication that has reported on the same information.
When I need to validate something fast, this combination of a quick prompt and two trusted sources usually takes less than 30 seconds — and gives me far more confidence in what I’m reading.
Always ask one of these three follow-ups to fix ChatGPT’s common weaknesses
As a general rule, I realize that ChatGPT is generally good at supplying definitions and explaining general concepts that delve into how things work. But what the chatbot is not so efficient at is giving you all the factual details connected to recent events, breaking down details attached to more niche topics, and listing proper attributions for their data.
To keep ChatGPT in check and avoid using phrases that sound real but aren’t real, I utilize these follow-up prompts to push it towards finding proper sources I can look up myself:
- Where did you get this statistic? [copied and pasted statistic]
- Can you cite a source for this information? [copied and pasted sentence]
- Is this data based on a real study? [copied and pasted data]
Using any of these three questions helps me get all the background information I need to look up if I suspect ChatGPT is giving me fake quotes, dubious statistics, and studies that have no clear place, time or origin.
And sometimes, I’ll even go as far as copy a specific sentence verbatim that looks off to me from ChatGPT and paste it into Google to search for details that back up its legitimacy. If nothing legit pops up, I know to ignore that ChatGPT answer altogether.
With another 30 seconds to spare, I also abide by this method for certain situations to get the most reliable answers I need.
The takeaway
ChatGPT and other chatbots like it, unreliable. They’re incredibly useful in the right context.
I use them all the time for brainstorming, getting recommendations tied to my hobbies, building playlists or exploring offbeat questions just to see what they come up with.
But when I’m relying on ChatGPT for anything more substantive, I treat it differently. I push it to show its sources, then verify everything myself — quickly cross-checking key details to make sure the information holds up. And that's the balance. ChatGPT brings the speed. I bring the scrutiny.
Together, it’s a system that lets me move fast without sacrificing accuracy — and it’s one anyone can use to level up their research.
Follow Tom's Guide on Google News and add us as a preferred source to get our up-to-date news, analysis, and reviews in your feeds. Subscribe to Tom's Guide on YouTube and follow us on TikTok.
More from Tom’s Guide
- Most people use AI like Google — but ‘Architects’ use this 3-step shift instead
- I tested ChatGPT-5.4 vs Claude Opus 4.6 — is the $20 upgrade worth it?
- I stopped using one AI for everything: This 2-tool system reclaimed 10 hours of my work week
You must confirm your public display name before commenting
Please logout and then login again, you will then be prompted to enter your display name.
