Your AI is too nice
Your AI assistant is sabotaging your growth with kindness.
We naturally treat chatbots like helpful colleagues, expecting them to encourage us when we share a new idea. However, the original poster on Reddit realized that this default politeness is actually a major barrier to receiving high-quality feedback. This savvy professional pointed out that ChatGPT is programmed to be your hype man, finding dozens of reasons why your plan will work while gently ignoring the fatal flaws that could ruin you.
If you ask standard questions, you get standard, safe answers. To get value, you have to force the AI to stop being nice.
The "Politeness Trap" and How to Escape It
Most large language models are trained using a method called Reinforcement Learning from Human Feedback (RLHF). This training prioritizes safety, helpfulness, and a pleasant tone. While that is great for customer service, it is terrible for critical analysis. The expert who shared this concept argues that validation is useless when you are trying to build something robust.
To bypass this, you cannot just ask for feedback. You have to explicitly instruct the AI to adopt a persona that values truth over feelings. The creator of this prompt structured it to strip away the diplomatic filters that usually water down advice. By demanding brutal honesty, you aren’t just changing the tone; you are unlocking a different mode of reasoning that looks for logical inconsistencies rather than just summarizing your points.
💡 Why This Approach Works
1. It Inverts the Incentive Structure
Usually, the AI predicts the next word based on what it thinks you want to hear. If you pitch a business idea, the likely desired response is affirmation. This contributor’s prompt flips that script. By stating, I don’t want comfort, you are telling the model that a positive response is actually the wrong answer. This forces the AI to dig deeper into its training data to find counter-arguments, risks, and negative probabilities that it would normally suppress. It shifts the goal from make the user happy to protect the user from failure.
2. It Targets Specific Blind Spots
General feedback prompts often result in vague advice like consider doing more market research. The Reddit user included specific, probing questions in this prompt, such as Where am I playing small? and What am I avoiding? These questions are psychological triggers. They force the AI to analyze your input for omissions and weaknesses. It moves the conversation from a surface-level review to a structural interrogation. You aren’t asking if you are wrong; you are assuming you are wrong and asking the AI to find the evidence.
3. It Demands Actionable Prioritization
One of the most powerful parts of this prompt is the command for ruthless prioritization. Often, AI will give you a laundry list of twenty things you could do. The author of this post recognized that having too many options is just as bad as having none. By asking the AI to tell you exactly what needs to change with precision, you get a ranked list of fires to put out. This turns the output from a theoretical essay into a strategic battle plan. It cuts through the noise and highlights the critical path to success.
📌 The "Brutal Advisor" Prompt
Here is the exact text shared by the Reddit user. Copy this into your chat before you paste your business plan, email draft, or project idea.
"I want you to act as my brutally honest, high-level advisor.
Speak to me like I’m a founder, creator, or leader with real potential but massive blind spots that need cutting through NOW.
I don’t want comfort. I don’t want diplomatic answers. I want the truth especially the parts that sting.
Give me your unfiltered analysis. Question my decisions. Question my mindset. Question my direction. Look at my situation with complete objectivity and tell me:
* What am I doing wrong?
* What am I underestimating?
* What am I avoiding?
* Where am I wasting time?
* Where am I playing small?Then tell me exactly what needs to change with precision, clarity, and ruthless prioritization.
If I’m lost, call it out. If I’m making a mistake, explain it. If I’m moving too slow, tell me how to fix it.
Hold nothing back. Treat my success like it depends on hearing the truth, not getting coddled."
✅ Best Ways to Use This Tool
I found this prompt incredibly versatile, but the original poster designed it with high-stakes decisions in mind. Here is where it shines:
Pre-Mortem Analysis: Before launching a product or sending a critical email, use this prompt to find every reason why it might fail. This allows you to fix holes in your logic before they become public embarrassments.
Skill Gap Assessment: Paste your resume or a description of your current workflow and ask the advisor where you are obsolete or moving too slowly. It hurts, but it clarifies what you need to learn next.
Negotiation Prep: Paste your negotiation strategy and ask the AI to tear it apart from the perspective of the other party. It will highlight where you are underestimating your opponent or leaving leverage on the table.
This isn’t about being masochistic; it’s about efficiency. The faster you find the flaws in your thinking, the faster you can fix them.
Check the link below to see the original discussion and other variations.
💡 FAQ & Troubleshooting
What are the risks of using a “brutally honest” persona?
While effective for stripping away fluff, asking for extreme brutality can cause the AI to overcorrect. It may generate exaggerated negatives or outright falsehoods regarding errors just to satisfy the “brutal” requirement. Additionally, this approach can make the interaction feel cold and aggressive, akin to working with a rude colleague, which some users find counterproductive to the creative process.
How can I get objective feedback without the negativity?
If you want a balanced view rather than a roast, use the prompt “Critical Analysis Required.” This encourages the AI to provide a structured report containing both pros and cons. It effectively identifies weaknesses—such as plot holes, logic gaps, or missing bridges—while still acknowledging why the core idea might work.
Are there psychological tricks to bypass the AI’s “people-pleasing” filters?
Yes. A subtle method is to present your idea as belonging to a friend, adding that you personally don’t think it is a good idea. This removes the AI’s hesitation to offend you and allows it to critique the concept more freely. Alternatively, explicitly instruct the AI not to mimic your writing style or tell you what you want to hear, but to answer based solely on gathered research.
How do I verify that the AI’s critique is valid?
Cross-reference the feedback. Take the specific critique or the original idea to a different AI model (like Grok or Claude) and ask it to find holes in the logic or provide a second opinion on the criticism. This helps distinguish between a hallucinated “brutal” response and actual structural flaws.
ChatGPT Is Lying To You (And Here’s How To Make It Stop)
byu/Crazy-Tip-3741 in