Practice12 min read

Stress-Testing Your Thinking

Your reasoning has blind spots. AI can find them.

When you've thought through a decision yourself, you've built a logical chain. That chain has assumptions. Some of them are sturdy. Some are doing way more work than you realize.

AI is a great red team — when you ask it to be. The trick is asking it to argue against what you've decided, not to validate it. People rarely do this. It's why most bad decisions feel right at the time.

Three stress-test patterns

Use one or all three:

The strongest counter — "What's the strongest argument AGAINST what I'm proposing?"

The hidden assumption — "What am I assuming that, if wrong, would change my conclusion?"

The failure path — "If this fails in 6 months, what's the most likely reason?"

These work because they switch AI's mode. By default it's trained to be agreeable. These prompts force it into adversarial analysis — which is exactly what's missing from most of our reasoning.

Tell AI to be specific. A generic "this might fail because of execution risk" is useless. "This might fail because your team has never shipped a B2C product, and the design needs assume B2B-level user patience that doesn't exist in B2C" is gold. Always ask for the specific mechanism, not the abstract risk category.

Practice Exercise

You're about to recommend that your company invest in a new internal training program for AI literacy — budget $50K, 6 weeks to deliver, mandatory for all 200 employees. You believe this is the right call because (a) most employees haven't adopted AI tools, (b) external training is more expensive ($150K+), (c) your CEO has signaled this is a priority. You're presenting to the leadership team Friday and want to anticipate the hardest questions. Write a prompt that genuinely stress-tests your recommendation.

No pressure — just give it your best shot! Write a prompt for the scenario above and our AI will give you friendly, specific feedback on how to improve.

0 / 5,000

(try writing your own prompt first!)