ChatGPT simply works as promised. It is helping us to summarize articles, generate images, and it will soon be creating videos for us.
Open AI has made using ChatGPT so intuitive, that many of us don’t give a second thought to our prompts and the responses we get. And therein lies the problem.
If you're using ChatGPT regularly but suspect it could be delivering more value, you might be making common mistakes that are easy to fix.
From choosing the wrong model to failing to properly structure your prompts, small adjustments can significantly improve your AI interactions.
Here are seven common mistakes people make when using ChatGPT, and how to fix them).
1. Not providing enough context
ChatGPT is smart. Sure, it still hallucinates from time to time (that’s when it makes things up), and you should always fact-check its outputs. But ask it for a synopsis of a book, and chances are you’ll get a solid summary.
That’s already better than I could manage—especially given the growing stack of books I mean to read but never quite get around to. Still, despite ChatGPT’s intelligence, people often assume it instantly knows two things: a) what they’re talking about, and b) what they want it to do.
In reality, most users give ChatGPT too little to work with, not too much. So don’t hold back—go into specific details of what it is you want to get out of your interaction with the chatbot. A simple way to do this is ask ChatGPT to assume a specific role.
This could be asking ChatGPT to reply as though it were your personal coach, or answer from the perspective of a CEO. Similarly, you can specify who the target audience for the reply should be.
You can also tell it who the response is for. Maybe you want the concept of black holes explained like you're five, or need a jargon-packed bullet-point list to drop into tomorrow's presentation.
Give it context — and suddenly, ChatGPT starts giving you exactly what you need.
2. Not using ChatGPT to help itself
It’s generally easier to start writing an essay when you’ve already got some text in front of you, rather than when you’re staring at a blank page. The same applies to prompts.
Now that we’ve discussed the importance of providing context, it’s time to leverage ChatGPT’s own capabilities to help us get there. A detailed prompt is more likely to help you get what you need than a vague one, but also requires more time and effort to produce.
Before sending your prompt to ChatGPT, spend a few extra seconds and ask it to create or refine the first draft of your prompt. Instead of asking ChatGPT to generate a cat image, ask it for a detailed prompt you can reuse or adapt.
It may add details like the location, a description of its fur, and specifics about the cat's pose that you may not have considered.
Ask ChatGPT to write its own prompts. The results may surprise or inspire you!
3. Not training ChatGPT
ChatGPT took the world by storm because it offers everyone a helping hand. That’s great for straightforward prompts when general knowledge suffices, but solely relying on ChatGPT’s standard training set can be limiting at times.
If you want to elevate your ChatGPT experience, getting a tailor-made answer to your request is one of the quickest ways to do so.
Don’t just take my word for it. Jensen Huang, CEO of Nvidia—the company behind the chips used to train ChatGPT—says he uses the tool to help write his speeches.
“I give it a basic outline, upload some PDFs of my past talks, and have it write the first draft,” Huang explained.
If ChatGPT’s responses feel too generic, try giving it information it wouldn’t otherwise know—or show it examples of what you're after. You can upload multiple files at once, even if they’re in a zipped folder.
4. Not tweaking your request
Another common mistake people make when using ChatGPT is to accept the first answer they receive.
Maybe the prompt itself was too vague. Or perhaps ChatGPT’s answer lacked the depth you were hoping for. Either way, it’s good to keep in mind that you don’t have to accept the chatbot’s first response without refining things.
Luckily there’s an easy fix. Simply ask follow-up questions or ask for the response to be more concise.
What your next prompt should be depends on your goals and what ChatGPT’s first output was. Have a back-and-forth conversation about it and see if you can improve on the first result.
If you’re in a hurry, try out the simple follow-up prompt: Are you sure?
5. Assuming it knows exactly what you want
So you’ve crafted the perfect prompt and hit enter. But the result is still not what you were expecting.
Perhaps the prompt sounds clear to you because you already know what it should mean. But that’s no guarantee that ChatGPT is on the same page as you.
Instead of assuming that ChatGPT grasps your goals, try asking it if it has all the context and information it needs. Invite ChatGPT to ask clarifying questions about your prompt to make sure it fully understands what you want.
You can even steal this follow-up prompt to use next time: Before answering, ask me any questions about my prompt that may make things clearer for you or that can provide you with more information to help formulate your response.
6. Using only one model
With most of us constantly being in a hurry, it’s no surprise that we simply use whichever ChatGPT model happens to be selected at any given time.
If you feel like you could be getting more out of your interactions with ChatGPT, there may be a model that’s optimized to handle your request in a better way.
For example, GPT-4.5 is great at producing clear writing that’s easy to digest. If you need a helping hand with producing code or solving a math problem, then maybe try o3-mini-high which has been built for exactly these purposes.
ChatGPT’s models may look and feel quite similar but there are definitely cases in which the answers they give to a question differ. It happened recently in one of my tests where GPT-4o and ChatGPT o-3 mini-high gave a different answer to a question testing their logic.
Experiment with different models. Feed them the same prompts and see which results you like best. In a nutshell, don’t limit yourself.
7. Caring too much about typos
I’m ending this list with a controversial tip. Most of the advice on the internet actually says you should focus on using correct grammar and spelling. But hear me out. Isn’t ChatGPT here to make our lives easier?
When texting a friend, we usually want a quick answer—not a lecture on whether we should’ve used “effect” or “affect.” Similarly, most prompts don’t need a full research report—though ChatGPT can certainly provide one if needed.
For simple ChatGPT prompts, if there’s a typo here and there or if you’ve added a word when you clearly meant another, everything will be fine. There’s no need to painstakingly review your simple prompts, crafting them to perfection.
Abbreviations also work fine. Otherwise, any time ChatGPT saves you would be spent by you overanalyzing the prompt box.
Upload that foreign menu and just ask, “Translate to En.” Accidentally say your skin got burned by the moon instead of the sun? ChatGPT will still understand.
Now that you've learned the 7 ChatGPT mistakes you might be making, why not take a look at some of our other AI articles?
ChatGPT can help you find the best haircut for your face shape and even set reminders for you. And if you're looking to get better results from ChatGPT, check out these 11 prompting tips.
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.
