News for hackers. Stuff that matters.

A comment on Will training AI be the next front in the “culture war”?. Parent comment

Re:Will training AI be the next front in the “culture war”? by tus666 on Monday April 24, 2023 @12:14AM (#35683588)
Open.AI and ChatGPT itself. Ask it. The conversations ChatGPT feed back into the model, it's one of the ways they improve it. In particular, if you can get it to admit it's wrong on something, and it verifies that internally, it won't make the same mistake again. It's self-correcting in that way.
[ Parent | on HN ]
Re:Will training AI be the next front in the “culture war”? by smoldesu on Monday April 24, 2023 @12:36AM (#35683704)
Can you cite something that doesn't have a disclaimer reading "May occasionally generate incorrect information" on this? It's news to me.
[ Parent | on HN ]