News for hackers. Stuff that matters.
Re:OpenAI's justification for why training data is fair use,... by acomjean on Thursday October 05, 2023 @12:59PM (#37781059)
I think in some ways LLMs have shown people want simple answers to their questions when they pose it to a search engine.

Stack Overflow was good because it had answers (one chosen as right) and a little bit of discussion about why and how and other considerations.

Much of the web is verbose to an extreme to get the search engines to rank it highly. People don't want that, they just want answers.

Re:OpenAI's justification for why training data is fair use,... by verdverm on Thursday October 05, 2023 @01:12PM (#37781220)
This is exactly right, and hopefully the LLMs can get better at adding in the

> discussion about why and how and other considerations

If(when) they knew when/how to ask better follow up questions, that will take it to the next level