I Asked ChatGPT To Analyse My Chess Game — And It’s Not Good Enough
However, it learned quickly and corrected itself. So that's consoling.
The Game
This is a blitz game (3 min with no increment) played on lichess.com by me (rated 1760 at the time of the game) against a much stronger opponent (rated 2200 at the time of the game)
One thing that you can notice from the below screenshot is the offset between my clock and my opponent’s. This is because he got a little too confident and opted to challenge me with half the clock.
The shaded, Grey and White, strip on the right is the win predictor, with the larger shaded color having more win probability.
The Opening
I am a fan of unorthodox openings, and while playing as white, this one is my favourite. “1.a4", also called the Ware opening, is named after Preston Ware, an American grandmaster of the 1880s.
Enter Chat GPT
Chat GPT got completely wrong about the opening. The conversation is as below (unedited):
Me: “Say something about ware opening in chess”
The opening in chess refers to the first few moves of a game, in which each player develops their pieces and establishes a foothold on the board. The opening is a crucial stage of the game, as it sets the stage for the rest of the game and can have a major impact on the outcome.
There are many different openings in chess, each with its own unique characteristics and strategies. Some openings are designed to control the center of the board, while others aim to occupy key squares or build a strong pawn structure.
It is important for a chess player to choose an opening that fits their style and strengths, as well as take into account their opponent’s style and the specific characteristics of the position. Choosing the right opening can give a player a significant advantage and set the stage for a successful game.
Me: “Sorry you are wrong. I asked about the “ware” opening”
I apologize for the misunderstanding. The “ware” opening is not a well-known or commonly used opening in chess. I am unable to provide any information about this particular opening.
It is possible that you may have misspelled the name of the opening, or that you are referring to a different game or activity. If you could provide more context or clarification, I would be happy to try to assist you further.
Me: “The ware opening is white playing a4 as the first move”
The “Ware” opening is a chess opening that is named after American chess master Norman T. Whitaker, who was known for using this opening in his games. It is characterized by the move 1.a4, which is the first move for White.
The move 1.a4 is an unusual and somewhat unexpected move, as it does not control the center of the board or develop any pieces. Instead, it aims to prevent Black from playing the Sicilian Defense or other openings that involve …b5.
The Ware opening is generally considered to be a somewhat unorthodox and risky opening, as it allows Black to occupy the center of the board with their pawns and can lead to an asymmetrical pawn structure. It is not a particularly popular choice at the highest levels of chess, but it can be used as a surprise weapon or to throw off an opponent’s preparation.
Me: “How is Ware named after Whitaker? Have you lost it ?”
I apologize if my previous message was unclear. The “Ware” opening is named after American chess master Norman T. Whitaker, who was known for using this opening in his games. It is sometimes referred to as the “Whitaker Attack” or the “Whitaker Defense” in his honor.
Norman T. Whitaker was a chess master and inventor who was active in the early 20th century. He was a noted specialist in blindfold chess and was known for his unorthodox and aggressive style of play. The Ware opening is one of the openings that he was known for using in his games.
I apologize for any confusion in my previous message. I hope this clears up the misunderstanding. Please let me know if you have any further questions or if there is anything else I can help with.
Me: “Isn’t it named after Preston Ware”
I apologize for the mistake in my previous messages. You are correct that the “Ware” opening is named after Preston Ware, not Norman T. Whitaker.
Preston Ware was an American chess player and chess writer who was active in the late 19th and early 20th centuries. He was known for his aggressive and attacking style of play and was particularly fond of the Ware opening, which he named after himself. The Ware opening is a chess opening characterized by the move 1.a4, which was first played by Ware in a tournament game in New York in 1878.
I apologize for any confusion caused by my previous messages. Thank you for correcting my mistake. Please let me know if you have any further questions or if there is anything else I can help with.
The Middle Game
As you can see on the clock, my opponent has literally consumed no time for his first two steps. They are all routine ones, so I was not expecting anything different.
By this moment, being on the time advantage I have used up 13 seconds while my opponent has used only 1 second for all of these 5–6 moves. As you can see on the win predictor on the right, there is no significant advantage to any of us.
After 4–5 crazy moves from both sides, here’s how it is, I have used up 20 seconds while my opponent has used only 2 seconds, and the game is still dead on balance.
The End Game
The game is completely in my control by this time, had I taken the predicted move for the queen. You can see what I did in the next screenshot.
Yeah, you saw it right. I missed it. Not by much, but I missed it. This is as far as the win predictor goes. But in fact, I took a chance on this fast-moving player. He is moving so fast that he used only 4 seconds for his entire game.
And Voila!!! He falls for it. My opponent missed it this time, and unlike me, he missed it big time. If you haven’t seen the catch, don’t be disappointed. It’s not an easy find for first-timers.
And that’s the move, Qh5#. It’s checkmate. I defeated a player who is rated higher than me by nearly 500 rating points. My game made me proud, very proud, but do you know who has let me down again? — ChatGPT
Wrong again - This time on the Game analysis
ChatGPT completely got this game wrong. See how it unfolded:
Me: “Can you analyse ‘https://lichess.org/QKoGPmPQ’"
Certainly! Here is my analysis of the game at the provided link:
[Analysis of “https://lichess.org/QKoGPmPQ"]
Overall, it was a well-played game by both sides. White was able to get a small advantage in the opening by controlling the center of the board and developing their pieces efficiently. Black did a good job of defending and counterattacking, eventually equalizing the position.
In the middle game, both sides made a number of mistakes. White could have kept the advantage by playing more aggressively and not allowing Black to trade pieces and simplify the position. Black also missed a few opportunities to take the initiative and put pressure on White.
In the endgame, White made a crucial mistake by allowing the exchange of rooks, which left their king exposed and gave Black a decisive advantage. Black was able to convert their extra material into a win by trapping the white king and promoting their pawn to a queen.
Overall, it was a closely contested game with both sides making mistakes. With better play, either side could have emerged as the winner.
Me: “Very bad analysis”
I apologize if my analysis was not helpful. As a machine, it is possible that I may overlook certain details or make mistakes in my evaluation of the game. If you have any specific suggestions or corrections, I would be happy to consider them and improve my analysis.
Me: “There is no exchange of rooks at all”
I apologize for the mistake in my analysis. You are correct that there was no exchange of rooks in the game. Thank you for pointing out the error. Please let me know if you have any further questions or if there is anything else I can help with.
Me: ”Analyse again”
Certainly! Here is my revised analysis of the game:
Overall, it was a well-played game by both sides. White was able to get a small advantage in the opening by controlling the center of the board and developing their pieces efficiently. Black did a good job of defending and counterattacking, eventually equalizing the position.
In the middle game, both sides made a number of mistakes. White could have kept the advantage by playing more aggressively and not allowing Black to trade pieces and simplify the position. Black also missed a few opportunities to take the initiative and put pressure on White.
In the endgame, White made a crucial mistake by allowing Black to exchange their bishop for a knight and gain a strong passed pawn on the queenside. This gave Black a decisive advantage, and they were able to convert their extra material into a win by trapping the white king and promoting their pawn to a queen.
Overall, it was a closely contested game with both sides making mistakes. With better play, either side could have emerged as the winner. I hope this revised analysis is more accurate and helpful. Please let me know if you have any further questions or if there is anything else I can help with.
Me: “The extra material didn’t come into play at all”
I apologize for the mistake in my analysis. You are correct that the extra material did not come into play in the endgame. Thank you for pointing out the error.
It looks like Black was able to win the game by exploiting the weaknesses in White’s position, particularly the exposed king and the weak pawns on the queenside. Black’s bishop was also very strong in the endgame, and it played a crucial role in attacking the white king and supporting the promotion of the passed pawn.
I hope this revised analysis is more accurate and helpful. Please let me know if you have any further questions or if there is anything else I can help with.
Me: ”Yeah it is, but why did you get it wrong twice before this”
I apologize for any mistakes in my previous analyses. As a machine learning model, I rely on the data that I have been trained on to generate responses. Sometimes, my responses may be incorrect or inappropriate because I have not been trained on all of the possible variations and scenarios that can occur in a game of chess.
I am constantly learning and improving, and I appreciate your feedback and corrections. Please do not hesitate to let me know if you have any further questions or if there is anything else I can help with.
My take
I can clearly understand why ChatGPT got this wrong. It is because the game is very unorthodox and thus is not among the common training materials.
ChatGPT is very good, but only for what it's trained on and a little above and beyond. It can definitely not think or improvise beyond what it’s trained with.
If you like the content and would like to collaborate with me or hire me for a gig, please contact me at thewierdstoryteller@gmail.com. Please be mindful of the spelling, it’s weird, not weird.
