Field notes
Research updates, reflections, and dispatches from the intersection of AI and education.
We built Open Items — 34K+ CC-licensed assessment items with AI generation and LLM evaluation — as the first applied project on the CZI Learning Commons Knowledge Graph. Here's what we learned about building the open applied layer on top of open infrastructure.
Read more →Our paper on using large language models to estimate item difficulty has been accepted at AIED 2026. Here's what we found across 15+ models and 200 experimental conditions.
Read more →Education has a crisis of purpose. AI changes what students need to know — but nobody has the evidence to say what matters now. That's the gap we're filling.
Read more →