2026-09-27 16:14 UTC
@ahltorp@mastodon.nu @malte@radikal.social @emilymbender@dair-community.social
There is a question posed in a W.S. Graham poem: ‘What is the language using us for?’
I think of that from time to time to remind myself that utterances have an existence beyond the person or, indeed, the machine that ‘speaks’ them, which makes it problematic to locate meaning in intent or originating intelligence. Yes, in conversation, humans cooperate in the creation and interpretation of meaning, but that meaning subsists in grammar and vocabulary that we must all learn. We also independently intepret meaning every time we read a text, without direct access to the intent or clarification of the author, relying only on the meaning that might be sustained by the words and their arrangement.
Replies (1)
-
@TiroTypeworks@typo.social 2026-09-27 16:15
@ahltorp@mastodon.nu @malte@radikal.social @emilymbender@dair-community.social A machine inserted into the transmission of language doesn’t have to be intelligent, it only needs to be able to replicate the grammar and vocabulary in ways that sustain meaningful interpretation, which the LLM models do remarkably well. In this respect, it feels like something of a red herring to even ask whether ‘AI’ is or could become intelligent, because in practical terms it hardly matters to either the purposes to which the models might be put or to the social, economic, environmental, and other threats that they pose.