> reinforcement learning from human feedback, which is the same method used in ChatGPT

Is this confirmed? I thought it was not so.