Resisting the Automation of Thought
We train machines to respond more effectively and efficiently. But we should first train ourselves to ask better questions. We must not allow AI to think for us, and certainly not to preformulate the questions we are allowed to ask. In this age, critical thinking begins with the humility to recognize when an algorithm has already narrowed our field of vision, and with the courage to reopen the framework.
The following are 7 practical habits to keep your questions ahead of the answers that an AI can give
Enhancing Critical Thinking with AI (made with napkin AI)
1 — Pre-mortem the Prompt
Before you type anything into an AI tool, write down the most critical thing you want to understand. Then ask: “If the system gives me a perfect-looking paragraph, how could it still mislead me?” This one-sentence pre-mortem exposes hidden assumptions you’re about to bake into the prompt.
2 — Run the “Five Whys” on Yourself
After you get an AI response, interrogate your satisfaction: “Why does this feel complete?” Repeat the why-chain five times, usually by the fourth “why,” you surface a value judgment or political premise that the AI mirrors back at you.
3 — Switch the Frame, Fast
Take any AI answer and force it into three alien frames:
• Historical (How would a 1920s labor organizer read this?)
• Adversarial (How would a direct competitor exploit this claim?)
• Marginal (Whose voice is least audible in this summary?)
Ten minutes of frame-switching inoculates you against monoculture.
4 — Build a “Question Bank,” Not a Knowledge Base
Keep a living document that stores only questions — never answers. Every time an AI gives you a datum, translate it into a new question and append it. Over months, you’ll own a map of your evolving ignorance, which is more valuable than any static set of facts.
5 — Practice the 3-Layer Source Audit
• Layer 1: Ask the AI for its sources and confidence scores.
• Layer 2: Pick the top two sources and read them directly, ignoring the AI summary.
• Layer 3: Use the bibliography of those sources to find one paper that the AI did not mention.
This triple jump keeps the machine’s attentional lens from becoming yours.
6 — Schedule “Slow Thinking” Sprints
Once a week, block 25 minutes of no-screen time to focus on pen and paper. Pick one AI-generated statement and try to reconstruct the chain of evidence without looking it up. The physical slowness of handwriting forces gaps and contradictions to the surface.
7 — Teach the Machine Your Doubt
End every session by feeding the AI your top remaining uncertainty: “I’m still not sure how X interacts with Y under condition Z.” Save its reply, but also write your 100-word counter-scenario. Comparing the two side-by-side trains, you will spot where the machine’s imagination ends and yours must begin.
Remember: an answer is a product; a question is a process. Train the process daily.
This might be of interest:
You can read an excerpt from the book “Critical Thinking is Your Superpower” here:
