People repairing chatgpt replies with additional prompts is reinforcement learning training data. "Reinforcement learning", just like any term used by AI resea

People repairing chatgpt replies with additional prompts is reinforcement learning training data. "Reinforcement learning", just like any term used by AI researchers, is an extremely flexible, pseudo-psychological reskin of some pretty trivial stuff.