Here’s a bold thought: most of what we call “critical thinking” is really just a mix of skepticism and doubt. Real critical thinking is a structured process, not just a gut feeling about what feels wrong.
I just saw a post that completely reframes how to use AI for deeper analysis, and I was blown away. The original poster argues that instead of just asking simple questions, we can use specific frameworks to guide ChatGPT into being a powerful thinking partner. This isn’t about finding quick answers; it’s about using AI to challenge your own assumptions and build stronger arguments.
This approach uses prompts that are more like structured templates than simple questions. I think it’s an incredible way to approach complex problems.
🧠 Here are three of the best techniques the innovator shared:
📌 Deconstruct from the Ground Up
Instead of looking at a problem as a whole, this method forces you to break it down to its absolute core truths. The expert calls this “First-Principles Deconstruction.” It’s about stripping away all assumptions until you’re left with only what’s essential, then rebuilding a solution from there.
Here’s the prompt:
Act as a first-principles analyst. Decompose the problem: “{insert problem}” into its smallest assumptions, physical/financial/logical constraints, and causal mechanisms. For each assumption, evaluate its necessity (must-have vs convenience) and testability (how to falsify). Rebuild a solution from only the indispensable parts. Output in sections: Assumptions, Constraints, Mechanisms, Tests, Rebuild, Risks, Final Recommendation.
✅ Argue Against Yourself (Effectively)
This is my favorite one. Instead of creating a weak “strawman” version of the opposing argument, the “Steelman” technique makes you build the strongest possible counter-argument. The post’s author provides a prompt that helps you find the flaws in your own logic and then strengthen your original position. Awesome!
Use this one:
Act as an impartial philosopher. State my position: “{insert position}.” Identify the strongest possible opposing thesis that a highly intelligent critic would advance. Present the opponent’s best arguments, evidence, and counterexamples. Then respond by improving my position (not attacking a strawman) with revisions, concessions, and better evidence.
Output sections: My Thesis, Steelman Opposition, Rebuttal, Revisions to My Thesis, Updated Thesis.
💡 Plan for Failure Before You Start
The “Red Team Pre-Mortem” prompt is pure strategic genius. You tell the AI to assume your project has already failed and then work backward to identify the most likely causes. This contributor’s method helps you spot weaknesses and build contingency plans from day one.
Try this prompt:
Act as a red-team strategist. Assume the plan “{insert plan}” failed catastrophically in 12 months. List the top 10 plausible failure modes, early warning signals for each, severity/likelihood ratings, and concrete countermeasures. Conclude with a prioritized mitigation roadmap and a monitoring dashboard (leading indicators, thresholds, and actions).
These are just a few of the 25 prompts the person who shared it included. Go check out the full post to see the rest and grab the infographic!