Probably not optimal for it. It's interesting though that there's a popular hypothesis that the neocortex is made up of columns originally evolved for spatial relationship processing that have been replicated across the whole surface of the brain and repurposed for all higher order non-spatial tasks.
I don't see why that would be the case. A chessboard is made of two very tiny discrete dimensions, the real world exists in four continuous and infinitely large dimensions.
Comments
But I suppose that if you can train an llm to play chess, you can also train it to have spatial awareness.
Probably not optimal for it. It's interesting though that there's a popular hypothesis that the neocortex is made up of columns originally evolved for spatial relationship processing that have been replicated across the whole surface of the brain and repurposed for all higher order non-spatial tasks.
The key word here is "if".
https://www.linkedin.com/posts/robert-jr-caruso-23080180_ai-...
I don't see why that would be the case. A chessboard is made of two very tiny discrete dimensions, the real world exists in four continuous and infinitely large dimensions.