New study warns AI tools may narrow what users learn

A new study examining how people use ChatGPT and other AI tools to learn has found a potential problem: the models tend to produce narrower, less diverse sets of information compared to traditional methods like Google search. Researchers tested multiple AI models and found none matched the range of results users would get from a standard web search. The phenomenon, dubbed "knowledge collapse" by researchers, suggests that relying on AI for learning may inadvertently limit exposure to different viewpoints and information sources. The findings raise concerns about how widespread adoption of AI learning tools could affect education and professional development.