GPT‑5 feels confusing at first — these 5 expert tips make it feel smarter than ever
These smart tips will help you adapt and unlock the model's full potential

Now that we’ve had a full week with GPT-5, hopefully you’ve had the opportunity to play around and test it. Undoubtedly, you’ve noticed that it’s a bit different from GPT-4o. Millions of people have had mixed feelings about the new model; some love the added depth and intuition, while others feel frustrated by the sudden shift in behavior and tone.
I agree that GPT-5 takes a little getting used to. It doesn’t always respond the way you expect, especially if you’re used to the more literal style of GPT-4o. But once you start to understand how it “thinks” and adjust your prompting style accordingly, you may find it’s actually more powerful, creative and conversational than anything that came before.
Here are some of the more annoying features about GPT-5 and workarounds to make them less aggravating.
1. GPT-5 overassumes
GPT-5 is more autonomous and context-aware. Up until now we’ve had to overexplain our intent when prompting chatbots. Those days are gone because GPT-5 can infer more from less. You don’t always need super-specific instructions because the AI just “gets” it. You can now express your intent with fewer words.
Prompting tip: Use more natural, open-ended language as if you’re talking to a colleague you’re collaborating with and not a search engine. For example:
Try: “I’m prepping for a job interview. Can you help me sound more confident when talking about my weaknesses?”
Not: “Write 3 bullet points describing weaknesses in a professional tone for a job interview in human resources.”
Get instant access to breaking news, the hottest reviews, great deals and helpful tips.
2. Thinking mode kicks in unprompted
GPT-5 has thinking modes that activate based on your prompt. Don’t worry about the complexity of your question because the AI will automatically switch between a fast, surface-level mode and a slower, more analytical mode depending on the difficulty of your prompt.
When prompting, if you want deeper reasoning, ask it to “think step-by-step” or make a point to use prompts that trigger analysis (like comparisons, strategy or abstraction).
Try this: “Think step-by-step: how could a small business use AI to improve customer retention?”
3. GPT-5 won’t let things go
This model is better at retaining coherence across longer interactions. You may notice that GPT-5 asks you more questions to ensure it understands. At the same time, it handles follow-ups and evolves the conversations much more naturally than GPT-4o, so you can build on what you’ve already discussed. GPT-5 has an upgraded memory, so it retains even more.
For example, if you’re in a brainstorming session, try a variety of prompts.
Try: “Let’s keep going with that idea but make it funnier.” Then “Now shift the tone to be more emotional.”
4. GPT-5 is overconfident at times
This model is more multimodal and better at integrating different inputs. The latest upgrade means the AI can analyze images, charts and even tone of voice more effectively than ever before. For that reason, you can fully combine inputs more freely. If you’re using Voice or Vision, you’ll want to include more contextual cues now.
Try: “Here’s a photo of the inside of my fridge, can you help me meal plan?” or even “I’m feeling overwhelmed, talk to me like a friend.”
5. It expects you to think like AI
GPT-5 interprets vagueness or ambiguity more confidently. Remember the “blank space” prompt? That one works even better now because GPT-5 will often “fill in the blanks” more aggressively. Keep in mind that while this can be useful, the update can also be risky if you’re not clear.
If you don’t want the AI to make assumptions, you’ll need to be explicitly clear.
Try: “Don’t assume tone, wait for me to specify.” Or, “Ask me follow-up questions before writing.”
The takeaway
If GPT-5 has left you scratching your head, you’re not alone. The shift from GPT-4o to GPT-5 was a little rough for me, too, and I’m a power user. But here’s the thing: AI is evolving so quickly that we do ourselves a disservice by getting too attached to any particular model.
Once you understand the quirks, you can start working with GPT-5 instead of wrestling against it or longing for the days of GPT-4.
Acclimating yourself to all the chatbots, many of them free, will help you adapt and adopt the newest models in the future.
Follow Tom's Guide on Google News to get our up-to-date news, how-tos, and reviews in your feeds. Make sure to click the Follow button.
More from Tom's Guide
- I tested ChatGPT-5 vs Grok 4 with 9 prompts — and there's a clear winner
- I tested ChatGPT-5 vs Claude with 7 challenging prompts — here's the winner
- GPT-4 isn’t gone — here’s how to use it for free right now











Amanda Caswell is an award-winning journalist, bestselling YA author, and one of today’s leading voices in AI and technology. A celebrated contributor to various news outlets, her sharp insights and relatable storytelling have earned her a loyal readership. Amanda’s work has been recognized with prestigious honors, including outstanding contribution to media.
Known for her ability to bring clarity to even the most complex topics, Amanda seamlessly blends innovation and creativity, inspiring readers to embrace the power of AI and emerging technologies. As a certified prompt engineer, she continues to push the boundaries of how humans and AI can work together.
Beyond her journalism career, Amanda is a bestselling author of science fiction books for young readers, where she channels her passion for storytelling into inspiring the next generation. A long-distance runner and mom of three, Amanda’s writing reflects her authenticity, natural curiosity, and heartfelt connection to everyday life — making her not just a journalist, but a trusted guide in the ever-evolving world of technology.
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.