Calvin Chess, or how one AI breaks the rules as it sees fit...

[Ulfgeir]

Seems that they had a chessgame between ChatGPT and Stockfish (an actual chessbot). Let's just say that ChatGPT was "creative" in the interpretations of the rules...

 

[Davies]

Gee, it is almost like the rules of a game can be ignored at any time if you don't care about people refusing to play with you. Or aren't programmed to care about that.

 

[Mustrum_Ridcully]

I guess it's closer to Skynet than we'd like, inventing time travel and teleporting pieces in that don't exist into the fight. And

Spoiler

still loses

.

 

[Dannyalcatraz]

Seems that they had a chessgame between ChatGPT and Stockfish (an actual chessbot). Let's just say that ChatGPT was "creative" in the interpretations of the rules...

It’s like a high-level AD&D fighter going solo against a high-level AD&D Magic-User. And the fighter capitalized on the MU’s mistake.

 

[Clint_L]

I was entertained. A lot.

Also, this is why we keep telling students not to trust ChatGPT. It isn't sentient. It has no concept of reality. It is a language emulator. Incredibly powerful at certain tasks, but it needs a brain to guide it.

 

[Paul Farquhar]

It's playing Children of the Watch rules.