University of Maryland releases IdeaLens AI paper detector

Researchers at the University of Maryland have developed an AI detector called IdeaLens that looks past surface-level prose to analyze the underlying reasoning and narrative structure of academic papers. The tool is designed to help academic publishers filter out AI-generated slop while minimizing false positives for human researchers. When tested against large language models, IdeaLens mislabeled AI-generated text as human work just one percent of the time and correctly identified human-authored documents in about 95 percent of cases. Open article →