Developing critical thinking in the age of AI

Observing yourself before trying to “use AI properly”

• Take a starting snapshot: notice your natural reflexes when using AI (especially when time is tight or the answer sounds convincing)
• Identify situations where you tend to rush (pressure, complexity, reassuring answers, confirmation of your own ideas)
• Understand AI’s limits so you can better supervise it
• Go back to basics: AI doesn’t know your context, doesn’t deal with the consequences, and never signs the document
• Apply four simple supervision reflexes, tailored to common mistakes

Look at an AI response beyond the (good) first impression

• Understand why a response can be clear and well-written… yet still too general, superficial, or lacking context
• Spot the biases that can throw off your evaluation and build in a deliberate pause before judging
• Use 6 quality criteria to assess a response
• Choose the criteria based on the context

Send the right requests to AI (not on autopilot)

• Understand why, “without clear instructions,” AI tends to generalize, simplify, and fill in the gaps
• Structure your prompts using the ECRIS method
• Strengthen your critical intent: use AI to test an idea, explore limits, compare perspectives—rather than looking for a single “right answer”
• Add simple strategies (clarity constraints, target audience, comparison, challenging assumptions)

Take back control: I supervise AI—not the other way around

• Ask yourself the key question before using AI: is it really necessary in this situation?
• Be clear about what you delegate (carefully) and what you don’t:
• Remember this rule: the more sensitive, complex, and human-impacting a task is, the more your judgment is essential

Anchor a supervision commitment

• Put your personal guidelines into words: “Starting today… I commit to…”