Do large language models understand us?

Re:Do large language models understand us? by gorgoiler on Saturday December 18, 2021 @04:44AM (#29603217)

Instead of the computer/book telling the non-Chinese speaker what to write, what if it’s a human that speaks Chinese?

Does it matter then that the writer has no understanding of the language, when the backend clearly does?

(I think I get it actually, after writing these two questions, but I’ll post them anyway.)

[ Parent | on HN ]

Re:Do large language models understand us? by viraptor on Saturday December 18, 2021 @07:26AM (#29603976)

I think at "the writer has no understanding of the language, when the backend clearly does?" you circle back to the argument's main problem - is it possible for the non-human backend to "understand" the language, even though it behaves like it does.

[ Parent | on HN ]