How to tell if an AI tutor matches an expert teacher's judgment
A practical answer grounded in a runnable, published harness rather than opinion. Links to Blind expert-parity: can a model adjudicate like a credentialed examiner, and can you prove it?.
How to tell if an AI tutor matches an expert teacher’s judgment. The short answer: you test whether the AI helps students learn, not just whether the model sounds plausible. Josh Durey brings the education-domain judgment first, then uses runnable AI-quality harnesses to make that judgment measurable. This page walks the method and links to one you can point at your own system.
This is the method behind Blind expert-parity: can a model adjudicate like a credentialed examiner, and can you prove it?, which you can read and run.
1. Make the failure observable.
Pick the specific way this can go wrong and build the smallest input that triggers it. If you cannot make it fail on purpose, you cannot prove it works. A worked example: Can you trust the model that grades your content? Measuring when an AI judge waves through broken work.
2. Measure against a baseline, not a vibe.
Compare to a neutral control so the number means something. A score with nothing to compare it to is theater. A worked example: Measuring feedback integrity: a blind-solver that catches AI explanations leaking the answer.
3. Check it a second, independent way.
Re-run with a different model family or a held-out set. Agreement across independent checks is the only verdict worth trusting. A worked example: Same answer, different grade: measuring when an AI grader can’t hold a verdict.
The point is the overlap: deep classroom, curriculum, and assessment judgment made measurable with AI/eval tooling. Every claim here maps to a public, reproducible harness, not a slide. If you want this run against your own system, the method transfers directly.