Probes

Small protocols with visible methods, explicit evidence boundaries, and results that can be challenged.

2 public artifacts

• Probe 1 · 2026-08-09 · recognition

  Probe 01 — Recognition under correction

  A runnable dialogue probe that asks whether an AI tutor merely repairs an answer after a learner objects, or treats the objection as evidence that can change the shared account of the problem.

• Probe 2 · 2026-08-09 · inquiry

  Probe 02 — Suspended judgment

  A runnable dialogue probe for whether an AI tutor pauses when a learner’s difficulty supports several explanations, or prematurely commits to one diagnosis and teaches past the uncertainty.