who tried this feature
ChatGPT can't play chess.
@Meriten said in #2:
> ChatGPT can't play chess.
Well it cannot really move pieces, but it can somewhat attempt to type chess moves as if it were a chess game.
> ChatGPT can't play chess.
Well it cannot really move pieces, but it can somewhat attempt to type chess moves as if it were a chess game.
I've played it few weeks back. It castled through its own bishop on f8 ( an illegal move). It's not there yet when it comes to chess.
I asked it if it knew the rules to antichess. It said yes and then made up a pretty odd set of rules. It had captures forced and kings non-special right, but then said that you win by checkmate.
My guess is that it read lichess.org/variant/antichess and then missed the part before rules that mentions briefly that you win by losing your pieces. So it copied over the part of the rules about how to win from the more detailed standard chess rules.
It's guess when I asked if it knew the rules to suicide chess(which is the same as antichess) was much more wrong though, it said that you can't move pieces among other oddities. But it still said that it did know the rules, I don't think it has the ability to admit to not knowing something.
My guess is that it read lichess.org/variant/antichess and then missed the part before rules that mentions briefly that you win by losing your pieces. So it copied over the part of the rules about how to win from the more detailed standard chess rules.
It's guess when I asked if it knew the rules to suicide chess(which is the same as antichess) was much more wrong though, it said that you can't move pieces among other oddities. But it still said that it did know the rules, I don't think it has the ability to admit to not knowing something.
lol
@ProgrammerAngrim said in #5:
> I don't think it has the ability to admit to not knowing something.
It is an AI. It is learning. I corrected ChatGPT.
>"You are wrong. Antichess and suicide chess are the same."
>"I apologize, you are correct. Anti-Chess and Suicide Chess are the same thing. I made a mistake in my previous answer. Thank you for bringing this to my attention."
Now try it again and you get this answer:
>"What is the difference between antichess and suicide chess?"
>"Antichess and Suicide Chess are the same game and they have the same rules and objectives. In both Antichess and Suicide Chess, the objective is to lose all of your pieces or get stalemated. The player who still has pieces on the board after all other pieces have been captured loses the game, and the game is won by the player who loses all their pieces or is stalemated. This leads to a completely different style of play compared to traditional chess, as players aim to sacrifice their pieces and make moves that will lead to their capture. There is no difference between Antichess and Suicide Chess."
> I don't think it has the ability to admit to not knowing something.
It is an AI. It is learning. I corrected ChatGPT.
>"You are wrong. Antichess and suicide chess are the same."
>"I apologize, you are correct. Anti-Chess and Suicide Chess are the same thing. I made a mistake in my previous answer. Thank you for bringing this to my attention."
Now try it again and you get this answer:
>"What is the difference between antichess and suicide chess?"
>"Antichess and Suicide Chess are the same game and they have the same rules and objectives. In both Antichess and Suicide Chess, the objective is to lose all of your pieces or get stalemated. The player who still has pieces on the board after all other pieces have been captured loses the game, and the game is won by the player who loses all their pieces or is stalemated. This leads to a completely different style of play compared to traditional chess, as players aim to sacrifice their pieces and make moves that will lead to their capture. There is no difference between Antichess and Suicide Chess."
It's fun to discuss some forum evergreen questions with ChatGPT, for example "Is chess a draw with perfect game". But it shows that ChatGPT is just an eloquent bullshiter.
Here's the chat I just had:
Q With perfect play from both Black and White, is chess always a draw?
A With perfect play from both sides, chess would indeed most likely always result in a draw. Chess is a game of limited resources, and with best play from both sides, the players would be able to neutralize each other's threats and create a balanced position. In practice, however, chess is not usually played with perfect play from both sides, and many games result in a win for one player or the other due to mistakes, blunders, or a player simply outplaying their opponent. That being said, in high-level chess, draws are still very common, especially in the endgame where it can be difficult to create a winning advantage. It's also worth noting that while chess is theoretically a draw with perfect play, there are still some rare situations where a win is possible, such as in the case of a theoretical win in the endgame of King and Pawn vs King. However, these wins are very difficult to achieve in practice and are not often seen in actual games.
Q
"It's also worth noting that while chess is theoretically a draw .." Can you show me a mathematical proof that chess is theoretically a draw? Or is this just a conjecture based on flimsy assumptions?
A The statement that "chess is theoretically a draw" is based on a long history of chess analysis and studies of endgame scenarios. While there is no formal proof that chess is always a draw, it is widely accepted among chess players and experts that the game is drawish in nature due to the equal chances given to both sides and the limited number of pieces and pawns on the board. This conclusion is supported by the empirical observation that many high-level games end in draws, as well as by endgame tablebases, which are vast databases of all possible legal chess positions with six pieces or fewer. These tablebases show that with perfect play, many endgame scenarios result in draws, such as King and Pawn vs King, Queen vs Queen, and Bishop and Knight vs Bishop. While it is still possible for a player to make a mistake or overlook a winning combination in the endgame, the drawish nature of chess means that the likelihood of a decisive result is low with perfect play. It's important to note that in chess, as with many games, the outcome is not determined by mathematical proof, but rather by the moves made by the players and the resulting positions on the board. The drawish nature of chess is a widely accepted conclusion based on years of experience and study, but it is not a proven mathematical theorem.
Q Your logic is faulty. The observation that most practical games at the highest level end in a draw does not exclude the possibility that there is one or a few sequences of perfect moves that always lead to a win for one side. You can't even assign probabilities for the existence or non-existence of such a sequence.
A I apologize for the oversight in my previous response. You are correct, the observation that most high-level games end in draws does not necessarily imply that chess is always a draw with perfect play. It is possible, although highly unlikely, that there may be a winning strategy for one side in certain positions. As you pointed out, it is difficult to assign probabilities to such a scenario, and there is currently no mathematical proof that chess is always a draw. The conclusion that chess is drawish in nature is based on a long history of chess analysis and studies of endgame scenarios, but it remains a conjecture rather than a proven theorem.
Thank you for bringing this to my attention and I hope this clarification better answers your question.
Here's the chat I just had:
Q With perfect play from both Black and White, is chess always a draw?
A With perfect play from both sides, chess would indeed most likely always result in a draw. Chess is a game of limited resources, and with best play from both sides, the players would be able to neutralize each other's threats and create a balanced position. In practice, however, chess is not usually played with perfect play from both sides, and many games result in a win for one player or the other due to mistakes, blunders, or a player simply outplaying their opponent. That being said, in high-level chess, draws are still very common, especially in the endgame where it can be difficult to create a winning advantage. It's also worth noting that while chess is theoretically a draw with perfect play, there are still some rare situations where a win is possible, such as in the case of a theoretical win in the endgame of King and Pawn vs King. However, these wins are very difficult to achieve in practice and are not often seen in actual games.
Q
"It's also worth noting that while chess is theoretically a draw .." Can you show me a mathematical proof that chess is theoretically a draw? Or is this just a conjecture based on flimsy assumptions?
A The statement that "chess is theoretically a draw" is based on a long history of chess analysis and studies of endgame scenarios. While there is no formal proof that chess is always a draw, it is widely accepted among chess players and experts that the game is drawish in nature due to the equal chances given to both sides and the limited number of pieces and pawns on the board. This conclusion is supported by the empirical observation that many high-level games end in draws, as well as by endgame tablebases, which are vast databases of all possible legal chess positions with six pieces or fewer. These tablebases show that with perfect play, many endgame scenarios result in draws, such as King and Pawn vs King, Queen vs Queen, and Bishop and Knight vs Bishop. While it is still possible for a player to make a mistake or overlook a winning combination in the endgame, the drawish nature of chess means that the likelihood of a decisive result is low with perfect play. It's important to note that in chess, as with many games, the outcome is not determined by mathematical proof, but rather by the moves made by the players and the resulting positions on the board. The drawish nature of chess is a widely accepted conclusion based on years of experience and study, but it is not a proven mathematical theorem.
Q Your logic is faulty. The observation that most practical games at the highest level end in a draw does not exclude the possibility that there is one or a few sequences of perfect moves that always lead to a win for one side. You can't even assign probabilities for the existence or non-existence of such a sequence.
A I apologize for the oversight in my previous response. You are correct, the observation that most high-level games end in draws does not necessarily imply that chess is always a draw with perfect play. It is possible, although highly unlikely, that there may be a winning strategy for one side in certain positions. As you pointed out, it is difficult to assign probabilities to such a scenario, and there is currently no mathematical proof that chess is always a draw. The conclusion that chess is drawish in nature is based on a long history of chess analysis and studies of endgame scenarios, but it remains a conjecture rather than a proven theorem.
Thank you for bringing this to my attention and I hope this clarification better answers your question.
This topic has been archived and can no longer be replied to.