The Missing Pillar of AI Literacy

AI literacy today focuses on using AI well. But there’s a deeper risk: the better AI gets, the less we practice thinking for ourselves.

When answers come easy, we stop doing the hard work that builds real understanding:

• Cognitive offloading — letting AI think so we don’t have to
• Deskilling — losing abilities we no longer practice
• False confidence — mistaking smooth answers for real knowledge
• Blind spots — missing errors because we’ve stopped looking

The OECD’s AI Literacy Framework covers four areas: engaging with AI, creating with AI, managing AI, and designing AI. It’s solid.

But it focuses on using AI systems. It doesn’t protect the human skills that fade when AI is always available—curiosity, questioning, and knowing what you actually understand.

You can become "AI fluent" while losing the ability to think independently.

The current framework can’t fully protect:

• The drive to ask questions worth exploring
• The skill of forming questions that lead somewhere
• The habit of checking whether you truly understand
• The instinct to test answers before trusting them
• The discipline to work without AI when learning requires it

These aren’t extras. They’re the missing pillar.

We propose adding a fifth pillar: Questioning AI.

Not "how to get better answers from AI." How to stay a capable thinker when answers are everywhere.

This means managing your own mind—not just the tool. Knowing what to delegate, what to keep, what to verify, and what you can explain on your own.

AI doesn’t care if you understand. It won’t protect your curiosity. That’s your job.

You can’t teach this as a checklist. It’s a stance—habits practiced repeatedly:

• Friction by design — pause to clarify what you’re asking before getting answers
• Curiosity loops — notice, wonder, identify what you need to learn
• Verification traces — show how you checked, not just what you concluded
• Independence practice — try without AI first, then use it to refine
• Failure rehearsal — look for how answers could break or mislead
• Social calibration — share early, invite pushback, revise based on feedback

Current frameworks teach how to use AI. This pillar teaches when, why, and whether to use it—and how to stay sharp either way.

Without it, we risk mass deskilling: people who can operate AI fluently but can’t think without it.

Build curiosity and thinking habits first. Tool skills follow naturally.

The goal isn’t better AI users. It’s better thinkers in an AI-saturated world.