#LLM hallucinations

The AI Hype Index distills current chatbot trends: widespread everyday use, growing regulatory scrutiny, emerging transparency from OpenAI, and government adoption despite open questions.

'OpenAI research shows hallucinations are statistical consequences of pretraining and that binary benchmarking incentivizes overconfident guessing; changing evaluation to reward calibrated uncertainty can reduce hallucinations.'