So a few days ago I was running a little experiment out of my own curiosity, I was putting ChatGPT up against Stockfish to see how language models like ChatGPT handle that kind of stuff.
In the end I had to end the experiment because ChatGPT refused to make any legal moves and I just have one question, why?
Probably just read the T&C of ChatGPT or ask their support. You didn’t say what you were doing, so if it was questionable, it probably has some heuristic safeguards in place.
It was chess.
Not sure what this question has to do with the topic of this forum.
Yeah, my guess however is that I’m pretty sure it’s just that ChatGPT can’t always keep a accurate view of the board, so it’s most likely just making guesses.