And so it starts, I suspect we'll be seeing this issue more and more since it's easy to just get GPT to spit out some text. I believe that the true beneficiarie

And so it starts, I suspect we'll be seeing this issue more and more since it's easy to just get GPT to spit out some text. I believe that the true beneficiaries of LLMs are those that are experts in their fields. They can just read the output and deal with the inaccuracies. Does anyone know if training an LLM with just one type of data, law in this case, creates a more accurate output?