Re:Ask HN: What is the AI that Ray Kurzweil talks about in L...
by brutus1213 on Saturday September 17, 2022 @09:47PM
(#32883567)
Most modern language models seem to do that to an extent. Even github's co-pilot sort of does this (with code instead of books). The models can even synthesize, which starts getting really impressive. Check out Flamingo from Google, which is a VLM (vision-language model). There, it can answer questions about the content of new images and even incorporate new information.
[ Parent
| on HN ]
There are no comments on this comment yet.