In this video, Dylan Davis explains that AI models are typically trained to be helpful and polite, which often results in vague or “soft” feedback that avoids criticizing your ideas. To fix this, he introduces the “BRUTAL” method—a six-step framework designed to force AI into providing honest, critical, and actionable feedback.
The BRUTAL Method
- B – Begin Fresh: Use “Temporary Chat” or “Incognito” modes (available in ChatGPT, Claude, and Gemini) to disable the AI’s memory. This prevents the model from relying on your past preferences and patterns, ensuring a more objective response.
- R – Right Model: Different AIs have different biases. Davis suggests that Grok and DeepSeek are naturally more blunt, while Gemini, ChatGPT, and Claude lean toward being supportive. For high-stakes ideas, he recommends testing the same prompt across multiple models.
- U – Use a Critic Persona: Explicitly tell the AI to act as a specific type of critic. He suggests three levels of intensity:
- Devil’s Advocate: For general counterarguments.
- Red Teaming: To actively hunt for weaknesses and loopholes.
- Gordon Ramsay: For harsh, surgical, and blunt feedback (specifically requesting that it remain actionable).
- T – Third-Party Framing: Detach yourself from the idea. Since AI tries to protect the user’s ego, telling the AI the idea belongs to a “co-worker,” “competitor,” or “random person” encourages it to be more critical.
- A – Ask Specific Questions: Avoid vague prompts like “What do you think?” Instead, ask targeted questions such as:
- “What is the biggest financial risk here?”
- “If this fails in six months, what would be the most likely reason?”
- “What would a skeptical investor find wrong with this?”
- L – Leverage AI Against Itself: If the feedback is still too nice, ask the AI to grade its own response. Use a prompt asking the AI to “rate your previous feedback from 1-100 on how genuinely critical it was,” identify the weakest points of that critique, and rewrite it to be harsher.
Summary of Key Use Cases
This method is particularly useful for:
- Pitching: Hardening a proposal before showing it to a boss or client.
- Product Launches: Finding flaws in a campaign before spending money.
- Difficult Communications: Getting a third-party perspective on sensitive emails or negotiations.
- Major Commitments: Stress-testing legal or financial decisions.
The video concludes by suggesting that users can also add “Custom Instructions” to their AI settings to permanently prioritize bluntness and substance over compliments in every conversation.