The Limitations of Chat GPT

RLHF, which stands for reinforcement learning with human feedback, is incredibly powerful in my opinion. This is 100% reinforcement learning paradigm in which you have a human that instructs and corrects uninteresting things or irrelevant things with respect to the context. So it's the combination of these three things, in my opinion, that makes a difference with respect to what chat GPT, or the family of GPT models, could provide in terms of experience and of course usability.