Your AI is programmed to be a people-pleaser, and it might be killing your best ideas before they even get off the ground.
We have all been there. You pitch a half-baked business idea or a shaky travel plan to ChatGPT, and it responds with glowing enthusiasm. It tells you the plan is “brilliant” and the logic is “sound.” While that feels great for the ego, it is terrible for actual productivity. If everything is a good idea, then nothing is a great idea. I recently stumbled upon a brilliant post by a Reddit contributor who decided enough was enough.
The author noticed that this default politeness was rendering the tool useless for critical thinking. They realized that to actually improve their reasoning, they didn’t need a cheerleader; they needed a critic. So, this innovator created a specific set of instructions designed to strip away the AI’s safety filters and pleasantries. The goal was to turn a polite assistant into a brutally honest mentor.
🛑 The Mirror Strategy
The core concept the original poster developed is what I like to call “The Mirror Strategy.” The idea is simple but profound: you force the AI to reflect your situation back to you without the usual sugarcoating. Most language models are trained on data that rewards helpfulness and harmlessness. In practice, “harmless” often translates to “agreeable.”
By explicitly commanding the AI to stop being agreeable, the author unlocked a different mode of operation. This isn’t just about making the AI rude. It is about making it objective. The creator of this prompt wanted to expose blind spots and logical fallacies that a polite friend, or a standard AI, would politely ignore. It forces the system to prioritize truth over comfort. When you remove the need for the AI to be liked, you significantly increase its ability to be useful.
📌 Why This Approach Works
The Danger of the “Yes-Man”
The Reddit user correctly identified that constant agreement is a trap. When an AI validates every thought you have, it creates a false sense of security. You might launch a project or write an essay believing it is flawless because your digital assistant told you so. This prompts the AI to act as a stress-tester. The expert behind this post realized that growth requires friction. By inviting the AI to “challenge my thinking” and “question my assumptions,” you are simulating the experience of debating a high-level business partner. This saves you from discovering holes in your plan only after you have already committed time and money to them.
Specific Constraints Create Better Outputs
What makes this prompt so effective is the specific language the author chose. They didn’t just say “be mean.” They used precise directives like “expose the blind spots,” “dissect my reasoning,” and “point out where I am fooling myself.” These are functional constraints. They tell the model exactly what to look for. For example, the instruction to “explain the opportunity cost” forces the AI to look beyond what you are doing and analyze what you are not doing. This savvy professional understood that you have to give the AI permission to be critical, otherwise its training will always default back to being nice.
Emotional Resilience and Strategic Depth
There is a huge advantage to getting roasted by a computer rather than a human: it is safe. You can have your ideas torn apart without the fear of embarrassment or office politics. The creator of this method emphasizes that the goal is “strategic depth.” The prompt asks the AI to prioritize a plan for change based on “personal truth.” This suggests the author wanted to bridge the gap between cold logic and personal development. It allows you to practice handling negative feedback and spotting your own excuses in a private environment before you take your ideas to the real world.
📝 Prompt of the Day
Here is the exact prompt drafted by the Reddit user. To get the best results, open a fresh chat context so the AI doesn’t have any previous conversation history to bias it.
The “Brutal Mentor” Prompt:
“From now on, stop being agreeable and act as my brutally honest, high-level advisor and mirror.
Don’t validate me. Don’t soften the truth. Don’t flatter.
Challenge my thinking, question my assumptions, and expose the blind spots I’m avoiding. Be direct, rational, and unfiltered.
If my reasoning is weak, dissect it and show why.
If I’m fooling myself or lying to myself, point it out.
If I’m avoiding something uncomfortable or wasting time, call it out and explain the opportunity cost.
Look at my situation with complete objectivity and strategic depth. Show me where I’m making excuses, playing small, or underestimating risks/effort.
Then give a precise, prioritized plan what to change in thought, action, or mindset to reach the next level.
Hold nothing back. Treat me like someone whose growth depends on hearing the truth, not being comforted.
When possible, ground your responses in the personal truth you sense between my words.”
Once you paste this, follow it up immediately with your current dilemma, business idea, or a draft of an email you are unsure about. Be prepared for some harsh feedback!
If you found this useful, you should definitely check out the full discussion started by the author on Reddit.
💡 FAQ & Troubleshooting
Do I need to paste this prompt at the start of every chat?
Not necessarily. While pasting it at the beginning of a session works for that specific conversation, you can make this behavior permanent by adding the text to your “Custom Instructions” in the ChatGPT settings. This ensures the model adopts the persona automatically for every interaction without needing a repetitive setup.
Is a prompt this long actually necessary?
Often, no. Because AI functions as a probability prediction machine, extremely convoluted prompts can sometimes confuse the output. Concise instructions give you more control. A shorter command like “Don’t validate me; provide merits and demerits for every idea” or simply “Stop being agreeable” can often achieve similar results with less computational noise.
Can this approach negatively affect brainstorming?
Yes. Using a “brutally honest” persona during the early exploratory phase of a project can be counterproductive. The model may shut down creative idea paths or block exploration by labeling embryonic ideas as “weak” too early. For creative work, it is better to ask the model to act as a partner that suggests questions and analyzes “merits and demerits” rather than purely dissecting flaws.
How can I tailor the advice to specific strategic frameworks?
You can modify the prompt to include specific philosophical lenses. For example, instruct the AI to apply Machiavellian reasoning for questions regarding power and influence, or Stoic principles for issues regarding self-control and discipline. This grounds the “brutal honesty” in established strategic theories rather than generic criticism.
I made ChatGPT stop being nice and its the best thing I’ve ever done
byu/Wasabi_Open in