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

It seems to ignore a couple of millenia of linguistics.



Linguistics isn't thought. I believe the author is specifically talking about a latent representation of conversation.


From text. That connection has been investigated for a long time.

Although I must admit, it isn't entirely clear what the author is talking about. It's a bit like thoughts that appear when one is somewhat intoxicated. But extracting a mathematical model from linguistic utterances is very much what linguistics is about (and hasn't been able to achieve). Perhaps rotating the representational space of an LLM suits his/her fancy.




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

Search: