Researchers Say Chatbots 'Policing' Each Other Can Correct Some AI Hallucinations

Generative AI, the technology behind ChatGPT and Google's Gemini, has a "hallucination" problem. When given a prompt, the algorithms sometimes confidently spit out impossible gibberish and sometimes hilarious answers. When pushed, they often double down. This tendency to dream up solutions has already led to embarrassing public mishaps. In May, Google's experimental "AI Overviews"--these are […]