Powerful New AIs Are Making More Dangerous Hallucination Mistakes

AI models are boldly fabricating facts in 25% of chats – even trusted ones like ChatGPT invent 40% of their citations. The stakes are dangerously high in healthcare. Solutions exist.