The precondition for honest reflection
People only admit what they don't know when it's safe to — and that admission is the exact moment learning happens. Protect it architecturally, or it disappears.
Amy Edmondson’s research defines psychological safety as a shared belief that the team is safe for interpersonal risk-taking — that you can ask a question, admit a mistake, or say “I don’t know” without it being held against you. Teams that have it learn faster, because the information needed to improve actually surfaces. Google’s Project Aristotle later found it the strongest differentiator of its most effective teams.
“Psychological safety is a belief that one will not be punished or humiliated for speaking up with ideas, questions, concerns, or mistakes.”Amy Edmondson (1999)
AI makes this precondition load-bearing. A session transcript captures you not-knowing something and working it out. The moment people suspect those fumbling questions can reach a manager’s screen, they stop asking them — and the honest exploration that good work requires goes quiet. The data turns flattering, and the learning stops.
What to do on purpose
- Make the reflective space confidential by architecture, not by promise.
- Let coach transcripts and self-checks feed the person and their coach — never a dashboard.
- Let the learner see their own results first, and choose what, if anything, to share up.
- Log every access to anything that includes a person, where that person can read it.
In CARE this is Reflection, governed by the Embassy: the confidentiality is the feature. People will only examine what they don’t yet know if they are sure the examination is theirs.
Sources
- Amy C. Edmondson, “Psychological Safety and Learning Behavior in Work Teams,” Administrative Science Quarterly (1999); Google re:Work, Project Aristotle.