-
The thing with large-language model "AI" is that it doesn't have "oversight" or "focus". It's just fancy predictive-text: it puts together outputs based on the likelihood of one word appearing after another word within its training data. @RobDenBleyker/1828158131987231025Permalink On Internet Archive
On twitter.com
❤️ 2 Favorites
-
So if you tell a ChatGPT type application that it's wrong about something, it doesn't rethink or reevaluate. It just tries the words it knows in a different order.On Internet Archive