pull down to refresh

ChatGPT can already beat Magnus, all it takes is a very simple tool call to Stockfish.

isn't that sort of what the point of such a challenge would be? Not just mimic Stockfish, but actually use the everyday retail usage models of Chat (or Fable or Opus, because they apparently grasp chess better) and see when they can beat a human

But every day retail usage can already to tool calls I think. So you'd have to get pretty specific about the terms of the challenge, which is why I'm avoiding that market

reply

yea, true. Weird bet -- and it's too far into the future to tempt me, anyway.

reply