← All glossary entries

Reasoning

False Competence

Mistaking the model's fluency for your own expertise.

How it feels

The answer sounds clean enough that you feel smarter before you have checked whether you understand it.

At work

An analyst ships a confident market explanation assembled by AI, then cannot defend the causal chain under cross-examination.

Hidden cost

Reasoning: fluency replaces tested understanding.

You may have this if

Can you explain the answer with no AI, no notes, and one skeptical follow-up question?

Try this instead

Ask AI to quiz you on the answer, not improve the answer. If you fail, rebuild the reasoning yourself.

Team rule

Any AI-assisted recommendation needs a human-owned assumptions list before approval.

Related book concept

Loop layer and Socratic prompting.

Evidence notes

Direct

Microsoft Research (2025) reports lower self-reported critical thinking when knowledge workers place higher trust in AI.

Inference

False Competence names the subjective version of that trust gap.

Share card copy

False Competence: the model sounds fluent, so you feel fluent.

Open square card

30-second video script

Fluent language is not expertise. Before you trust an AI answer, ask: can I defend this against a smart objection? If not, you have borrowed competence, not built it.

Search surface

ChatGPT makes me feel smarter than I amAI false confidence at workAI fluency vs understanding

Related entries

Decisions Authority Borrowing Letting the model's confident tone carry a claim you have not verified. Reasoning Specious Certainty The calm feeling produced by polished language before evidence has earned it. Reasoning Rubber-Stamp Review Reviewing AI output for surface errors while accepting its frame.

Now measure the circuit

Find where this pattern is hitting you first.

The Anti-AI Brain Score maps your weakest circuit, then points you back to the practices that reload it.

Take the 2-minute score → Use this in your team