Gather round, fellow researchers. AI has become a valuable research companion, and it’s here to stay. But even good companions can lead you astray. What follows is no fiction: this issue features three true AI horror tales that made the news, of vanished work, phantom sources, and a creature that crept past peer review. Read them by candlelight, and take away the lesson each one teaches about using AI in research, so you won’t be the one haunted next.
Responsible AI
Machine Learning (ML) models are mathematical representations learned from data that enable machines to make predictions, classifications, or decisions. Without proper documentation, they function as opaque black boxes. In a fast-moving research landscape, model cards help ground innovation in clarity and trust. Think of them as READMEs for models with rich metadata that make ML models easier to discover, evaluate, reproduce, and share.