Run the Feynman gap check
Diagnose and repair fuzzy AI explanations using a repeat-back gap check.
The move: test the explanation by what the listener can repeat. A nod is not evidence of understanding. People often nod because the meeting needs to move on. The Feynman gap check asks for a short repeat-back: 'How would you explain this to someone on your team?' Their version shows what your words actually taught. What to listen for Listen for overtrust words: knows, decides, guarantees, learns everything, replaces people. Listen for undertrust words too: just autocomplete, random, useless. Each one points to a missing mechanism or missing boundary. How to repair Do not correct with a lecture. Repair one…
Sign up free — one personalized lesson every day, matched to your role and goals.
Already have an account? Sign in