Robot comment classifier

A developer built a classifier that detects robot-generated source code comments with roughly 80% accuracy. The model analyzes character, word, and part-of-speech frequencies, finding robots use more em dashes, semicolons, and full stops, while humans write more pronouns and TODO comments.