AI tools produce fewer hallucinations but more confidently wrong answers, study warns

Summary

AI tools are producing fewer obvious hallucinations but are increasingly generating inaccurate information presented with polished, hyper-confident language. This creates a dangerous dynamic where users trust AI outputs more and are less likely to catch costly mistakes, especially as reliance on AI grows for research, medical advice, and schoolwork. While obvious errors are easy to spot, the real concern is false answers that sound convincingly accurate. The article notes that AI proponents continue to advocate for keeping "a human in the loop" to mitigate these risks.