I mean they aren't large chess models. They can only do language tasks. They don't think, they predict words based on context and its similarity to the corpus they're trained on.

[–] Xavienth@lemmygrad.ml 34 points 1 week ago (2 children)

But if we just pump more language into them surely they will become sentient /s

[–] ALoafOfBread@lemmy.ml 17 points 1 week ago* (last edited 1 week ago) (1 children)

Disregarding the /s bc i want to rant

I guess if you described board states in language and got them to recognize chess board states from images (by describing them in language), and trained them on real games, you could probably make a really inefficient chess bot.

But that said, you could use an "agentic" model with an mcp to route queries about chess to an api that links the LLM to an actual chess bot.

Then it'd just be like going to the chess bot website and entering the board states to get the next move. No magic involved, just automated interaction with an api. The hype and fear and mysticism around llms bugs me. The concepts behind how they work aren't hard, just convoluted

[–] engineer@hexbear.net 10 points 1 week ago

This is really the future of LLMs, they're not going to directly replace workers like the marketers want us to believe. Instead they'll exist as very efficient interfaces between users and applications. Instead of applying all the correct headers to a word doc manually, you would use natural language to ask an LLM "Apply Headers to this document".

load more comments (17 replies)