It's just as the title says. I just read this article from Futurism and it made me wonder how Neuro would do if Vedal could connect her to the same system.
#ChatGPT lost at chess to an Atari. How would Neuro do?
proven hound
strange marsh
That has got to be the most insulting defeat in the history of chess
But whatever new chess system Vedal implemented for the most recent chess collab seems roughly 100 billion times better than ChatGPT while still being neuro making the decisions.
I assume this may be the clash of the ages
dreamy sable
tbh Neuro is most probably cheating, likely with a stockfish to generate candidate moves, and Neuro can choose from them (without the engine evaluation). at least, this is how I would implement Neuro’s chess functionality
(In reality, there’s probably some more sophisticated features for Neuro to pick her own moves, which makes her feel like a real little girl. The 1. h3 to 8. a3 sequence is not going to happen if Neuro was just picking from stockfish in any way)
The match result depends on the strength of the engine on the Atari 2600. Someone said around 1100 FIDE, which is stronger than most online chess players, so I think Neuro might struggle against it.
would be fun to farm views either way lmao
but really, it’s funny to see how human-like Neuro played. Some of the moves reminded of my first few games (e.g. the classic closed position on the kingside in the game against Filian)
drowsy moss
chatgpt also lost to a Twitch Chat despite being told it could cheat and make literally any move it wanted to