Hacker News new | past | comments | ask | show | jobs | submit login

They are the same problem, what do people think LLM is trained on.

"clickbait, "native content", clout-chasing, undisclosed sponsorship, and other such pathologies" is exactly the data set they used to train ChatGPT




Not once have I received a response from ChatGPT that meets any of those definitions.

…Perhaps the robots are better at truth than you are giving them credit for? :-)




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: