Llm as a judge
A practical answer grounded in a runnable, published harness rather than opinion. Links to Can you trust the model that grades your content? Measuring when an AI judge waves through broken work.
LLM as a judge. The short answer: you test whether the AI actually helps students learn, not just whether the model sounds plausible. I start with education-domain judgment, then use runnable AI-quality harnesses to make that judgment measurable. This page walks through the method and links to one you can run against your own system.
This is the method behind Can you trust the model that grades your content? Measuring when an AI judge waves through broken work, which you can read and run.
1. Make the failure observable.
I pick the specific way this can go wrong and build the smallest input that triggers it. If I can’t make it fail on purpose, I can’t prove it works. Worked example: Measuring feedback integrity: a blind-solver that catches AI explanations leaking the answer.
2. Measure against a baseline, not a vibe.
I compare to a neutral control so the number means something. A score with nothing to compare it to is theater. Worked example: Same answer, different grade: measuring when an AI grader can’t hold a verdict.
3. Check it a second, independent way.
I re-run with a different model family or a held-out set. Agreement across independent checks is the only verdict I trust. Worked example: Measuring feedback integrity: a blind-solver that catches AI explanations leaking the answer.
The point is the overlap. I bring deep classroom, curriculum, and assessment judgment, and I make it measurable with AI and 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.