Trust, But Verify: what a 194-citation audit found in AI-generated research

AI tools now generate research text complete with citations — but how many of those citations actually check out? In this talk I walk through what a 194-citation audit of AI-generated research found: which citations were real, which were hallucinated or misrepresented, and the patterns that separate trustworthy-looking output from output you can trust. The talk is aimed at students, researchers, and practitioners who use AI writing tools in their work. No single tool or vendor is the point of the talk; the point is a verification-first mindset: spot-check sources, trace claims back to primary material, and build review habits that catch AI-generated errors before they spread.

How to spot-check citations in AI-generated research; common patterns of hallucinated or misrepresented references; practical verification habits for anyone using AI writing tools.