Amanda’s AI Lab
Amanda’s AI Lab In my new Tom’s Guide column, I test the latest AI tools, features, and trends to see which ones are actually worth your time. We’ll explain what works and what doesn’t.
I’m the type of person who keeps a notebook full of ideas. Some are good, some are useless, and some I haven’t even thought about again since I wrote them down. I also keep notes on my phone and have sticky notes scattered around the office. It’s always a torrent of low-grade ideas.
That’s why I created prompts to help you bring your ideas back to reality and uncover weaknesses along the way. You can use it for almost any idea, and even when you can’t think of anything at all. In other words, the calm after brainstorming.
I stumbled upon this after asking ChatGPT multiple times to “improve” my idea, and I got polite and glossy feedback that made me feel like my thinking was smarter than it actually was. This model rephrases my half-baked logic in cleaner language, adds some encouraging transitions, and makes me feel like a genius. But the core problem still existed. Even after all this polishing, I can no longer see it.
Prompts to stress test your thinking
ChatGPT is great at encouraging, and sometimes that’s exactly what we need. But when it comes to big ideas, encouragement alone isn’t enough. We need resistance, real pushback, and tools that expose our shortcomings rather than polish them.
This is where what I call the Gravity prompt comes into play. Rather than asking ChatGPT to brainstorm, expand, or “make this better,” do the opposite. It forces models to act like hostile critics whose sole job is to poke holes, bring blind spots to the surface, and challenge shaky logic.
At its core, the Gravity prompt tells ChatGPT to stop being friendly and become hostile. It is designed to identify flawed assumptions, point out inconsistencies in reasoning, highlight overlooked risks, pressure and test conclusions, and ultimately separate the mere ones. sound It’s best if you can actually withstand it.
In other words, it drags your thoughts into reality. If you’ve ever left a meeting confident that your pitch was solid, only to have a co-worker ruin the meeting with one question you hadn’t even considered, you know the feeling. The Gravity prompt is its colleague. However, they are available at 2 a.m. and are not held back for reasons of courtesy.
Tackling problems by questioning assumptions
This is the exact prompt I use in ChatGPT because it tends to be the most people-pleasing. However, you can use any chatbot.
The “gravity” prompt is: It acts like gravity on my ideas. Your job is to bring it back to reality. It attacks the weakest points in my reasoning, questions my assumptions, and reveals what I may have missed. Be tough, be specific, and don’t gloss over your feedback. [Insert your idea].
ChatGPT’s responses may surprise you, as you’re likely to receive a completely different response than what you’re used to: sharper, more skeptical, and far less flattering. That’s the point.
This prompt works well because most people use AI as a hype machine. We ask our models to hone, refine, and expand their thinking, and they happily oblige. ChatGPT is comfortable by design. I hope it’s helpful, but this structure usually means building on what you’ve said rather than tearing it apart. The result is a false sense of confidence. Your idea reads better, but its underlying logic hasn’t really been tested.
Completely invert dynamics
Rather than building upwards, this prompt actively pushes downwards. Ask the model to find problems rather than solutions and weaknesses rather than strengths. And that resistance is where true clarity emerges. When your idea survives the gravity test, that is, when you manage to deal with all the objections the model throws at you, you know it’s actually solid.
I think this is very refreshing because you get honest, real feedback and not what ChatGPT was designed to be.
It’s also an effective antidote to what researchers call “pandering” in LLMs (Large-Scale Language Models), the well-documented tendency of AI chatbots to agree with users even when they’re wrong. By explicitly telling the model to push back, you are essentially overriding its default behavior and unlocking more honest and useful interactions.
If you need to use the “Gravity” prompt
This technique is especially useful before making high-stakes decisions where you need to be sure of your thinking. This includes business pitches and investor materials, strategic plans and project proposals, article concepts and creative briefs, startup ideas and product concepts, controversial opinions you plan to express publicly, and anything that seems “too true” to be true.
When I say I use this all the time, I mean it. In my opinion, if your thinking breaks down under pressure, you’re better off knowing it in a chat window than in a boardroom.
After using this approach for a few months, the workflow I’ve settled on is:
- beginningWrite your ideas clearly in one or two focused paragraphs. Avoid bullet points, as the chatbot needs to understand your ideas as completely as possible. Otherwise, the chatbot will assume and add ideas. Therefore, we will explain the core arguments and concepts in plain language. This step alone forces useful clarity, since vague ideas breed vague criticism.
- Number 2Run through the “Gravity” prompt and let ChatGPT do its worst.
- thirdI read all criticisms and honestly evaluate which ones are legitimate. Not everything is like that. In some cases, models may misunderstand nuances or raise issues that don’t apply. But more often than I’d like to admit, it captures something true.
- 4thRevise your thoughts based on valid critiques and run through the Gravity prompt again. Repeat this cycle until you find few structural problems with the model, but no problems.
In the end, my arguments are tighter, my blind spots are smaller, and my confidence is based on more than atmosphere.
conclusion
If you want ChatGPT to make you feel smart, don’t use this prompt. But if you want to make your ideas smarter – stress test your thinking like a tough-minded coach or a skeptical investor – this is one of the most valuable tips I’ve found.
to follow Tom’s Guide to Google News and Add us as a preferred source Get the latest news, analysis, and reviews in your feed.
