“A Very Unlikely Chess Game”, Scott Alexander2020-01-06 (, ; backlinks; similar)⁠:

…Black is GPT-2. Its excuse [for this chess blunder] is that it’s a text prediction program with no concept of chess. As far as it knows, it’s trying to predict short alphanumeric strings like “e2e4” or “Nb7”. Nobody told it this represents a board game. It doesn’t even have a concept of 2D space that it could use to understand such a claim. But it still captured my rook! Embarrassing!…Last month, I asked him if he thought GPT-2 could play chess. I wondered if he could train it on a corpus of chess games written in standard notation (where, for example, e2e4 means “move the pawn at square e2 to square e4”). There are literally millions of games written up like this. GPT-2 would learn to predict the next string of text, which would correspond to the next move in the chess game. Then you would prompt it with a chessboard up to a certain point, and it would predict how the chess masters who had produced its training data would continue the game—ie make its next move using the same heuristics they would. Gwern handed the idea to his collaborator Shawn Presser, who had a working GPT-2 chess engine running within a week:…You can play against GPT-2 yourself by following the directions in the last tweet, though it won’t be much of a challenge for anyone better than I am.

…What does this imply? I’m not sure (and maybe it will imply more if someone manages to make it actually good). It was already weird to see something with no auditory qualia learn passable poetic meter. It’s even weirder to see something with no concept of space learn to play chess. Is any of this meaningful? How impressed should we be that the same AI can write poems, compose music, and play chess, without having been designed for any of those tasks? I still don’t know.

[Shawn comments on HN. See also the much later Noever et al 2020a/Noever et al 2020b/Toshniwal et al 2021 who do the exact same thing in applying GPT-2 to Go SGF/chess PGN games. Shawn Presser’s encoding of the data turns out to be equivalent to Decision Transformer.]