Devil's Advocate / Steelman

AI can get stuck agreeing with you — it’s trained to be helpful and pleasant. This recipe breaks that pattern. A Devil’s Advocate challenges your idea to find its weaknesses. A Steelman builds the strongest possible version of the opposing view. Both make your thinking sharper.

The Devil’s Advocate Recipe

I've decided to [your decision or plan]. I want you to play devil's advocate — push back hard on this idea. What are the strongest arguments against it? What could go wrong? What am I probably not considering? Don't soften it.

The Steelman Recipe

I believe [your position]. But I want to understand the other side as charitably as possible. Give me the strongest, most thoughtful version of the opposing argument — not a strawman, but the best case someone who disagrees could make. Then tell me where you think the strongest points of genuine disagreement lie.

Example — Deciding whether to freelance

I've decided to quit my job and go freelance as a graphic designer. I want you to play devil's advocate — push back hard on this idea. What are the strongest arguments against it? What financial, professional, or personal risks am I probably underweighting? What do most people get wrong about freelancing? Don't soften it.

New to AI? This recipe works best after you’ve already made a decision and want a gut-check. After reading the response, you can ask “Now give me the best counter-arguments to your objections” to stress-test the other side too.

🔁 Leftover Remixes

🌶️ Spicy: Run both at once: “Steelman my idea AND devil’s advocate it simultaneously. Show me the strongest case for and against, then tell me where the real crux of the disagreement lies.”

🧊 Mild: Just the headlines: “Give me the top 3 objections, each in one sentence. No explanation needed.”

💰 Budget: “What’s the cheapest, lowest-risk experiment I could run this week to test whether the main objection is actually true?”