AI expert Gary Marcus references HAI's study showing that LLM responses to medical questions highly vary and are often inaccurate.
