Writing

What actually holds up when you test it.

Essays on AI-generated content, quality measurement, and learning science in education technology — grounded in evidence, not vendor claims.

Accurate is not the same as right

AI can be accurate and still fail a student: wrong exam board, or a summary that skips the retrieval work revision depends on. Precision, not correctness, is the real product.

The work behind a quality claim

Most AI trust claims are a rubric with no evidence anyone agrees on it. I tested mine on six reviewers before trusting it. One criterion, tone, failed completely.

The Future of Learning is Forged, Not Fabricated

Quality content is built through expertise and rigorous process, not shortcuts. Using blacksmithing as the frame: durable learning resources require careful source selection, real craft, structural discipline, and feedback — the best content is forged, not fabricated.