4 ms·
>It’s just important to remember that language is effectively just pointing at things, and that the words we use are effectively arbitrary. I think the success
by bugglebeetle 3y ago
>It’s just important to remember that language is effectively just pointing at things, and that the words we use are effectively arbitrary.
I think the success of transformer-derived LLMs suggests that Wittgenstein was more correct about contextual “family resemblances” than the “pointing at things” school of language philosophy.
- scoofy 3y agoI'm a bit confused by your statement. I'm a bit ignorant of the actual math/logic for modern LLMs (I took Andrew Ng's course on machine learning, but that just gets you to basic neural network architecture). I'm also confused as to why you would put Wittgenstein in opposition to the concept of language as "pointing at things," as I would put him very much in that school. I'm a bit rusty, but I've been through the Philosophical Investigations a couple times, and I'm almost certain he's opposed to platonic forms.
- beezlebroxxxxxx 3y agoHe thinks some words rely on ostensive definitions, colour for example. Otherwise, though, his view of language can roughly be summarized as "meaning is use". You're correct to read him as opposed to platonic forms and ideas; however, he argues that language and meaning has its roots in behaviour, or as he puts it: "a form of life". So for him meaning is relative but not arbitrary. It's a crucial distinction.
- scoofy 3y agoI only wish I could have heard him lecture. His two books, while interesting, have always seemed, well, a bit odd in their structure (less so with PI). Beyond that, that people use some random notebooks like his published texts also seems odd.