A survey found that about two-thirds of the references cited by AI were either non-existent or incorrect.

hallucinations . A study using OpenAI's large-scale language model, GPT-4o , revealed how prone AI is to hallucinations when asked specialized ...