“OpenAI is using a combination of techniques, including reinforcement learning with human feedback (RLHF), retrieval, and search, to reduce model hallucinations and provide more factual outputs.”