lichess.org
Donate

The Use of Ai When Analysing and general advice/questions

midori-laser

Hello everyone,

I’ve recently been trying to take my chess more seriously, and as part of that I’ve started analysing games, both my own and those of top players. In doing so, I often find myself wondering why a particular move was played, especially at the highest level.
At times, I’ll lean on AI tools (such as Copilot or Gemini) to suggest possible reasons behind certain moves. However, I’m unsure how reliable these explanations are. It can feel challenging to be 40 moves into a game and still not be able to clearly explain a decision based solely on my own analysis.

So I have a few questions:

When you’re analysing a game and get stuck trying to understand a move, where do you go for deeper insight? Do you consult stronger players, specific resources, or something else?

With the growing number of AI tools available, is there one you’d recommend as being particularly strong for explaining chess ideas and decisions?

I’ve had a Chess.com subscription for nearly a year, but I’ve found the instructional content a bit underwhelming. While the puzzles are fine, there seems to be a surprising lack of reading material. I tend to enjoy learning more through reading than through videos. Are there any resources you’d recommend that fit that preference?

Addtionally, i have purchased Fishers' 60 Memorable Games and Cernev's Logical Chess. Chernev's book has been insightful and i would be keen for any recommendations of similar material

Given all that, I’d really appreciate any advice on how to better understand the reasoning behind moves during analysis.

Hello everyone, I’ve recently been trying to take my chess more seriously, and as part of that I’ve started analysing games, both my own and those of top players. In doing so, I often find myself wondering why a particular move was played, especially at the highest level. At times, I’ll lean on AI tools (such as Copilot or Gemini) to suggest possible reasons behind certain moves. However, I’m unsure how reliable these explanations are. It can feel challenging to be 40 moves into a game and still not be able to clearly explain a decision based solely on my own analysis. So I have a few questions: When you’re analysing a game and get stuck trying to understand a move, where do you go for deeper insight? Do you consult stronger players, specific resources, or something else? With the growing number of AI tools available, is there one you’d recommend as being particularly strong for explaining chess ideas and decisions? I’ve had a Chess.com subscription for nearly a year, but I’ve found the instructional content a bit underwhelming. While the puzzles are fine, there seems to be a surprising lack of reading material. I tend to enjoy learning more through reading than through videos. Are there any resources you’d recommend that fit that preference? Addtionally, i have purchased Fishers' 60 Memorable Games and Cernev's Logical Chess. Chernev's book has been insightful and i would be keen for any recommendations of similar material Given all that, I’d really appreciate any advice on how to better understand the reasoning behind moves during analysis.
<Comment deleted by user>
fin34601473braunpaul

All current AI tools lie big. Simple proof: Ask a current latge language model chat "How did you come up with this answer?". They can and will not tell you how they crunched numbers to synthesize a text, but they really did nothing else. There are better chess specific tools like decodechess, but it all boils down to the plain old tree of variants (that GM Kotov literally suggested in his book "Think like a Grandmaster" because a high school kid knows how to handle a tree diagram and this on its own is no proof that this method were of any use). I am not at all saying a tree diagram for chess moves were useless, I am only saying that the explanation I am referring to is not convincing. Sadly computers are so much better at running a tree up and down and that is the ultimate skill that makes them strong chess players even without any more modern software architecture.

All current AI tools lie big. Simple proof: Ask a current latge language model chat "How did you come up with this answer?". They can and will not tell you how they crunched numbers to synthesize a text, but they really did nothing else. There are better chess specific tools like decodechess, but it all boils down to the plain old tree of variants (that GM Kotov literally suggested in his book "Think like a Grandmaster" because a high school kid knows how to handle a tree diagram and this on its own is no proof that this method were of any use). I am not at all saying a tree diagram for chess moves were useless, I am only saying that the explanation I am referring to is not convincing. Sadly computers are so much better at running a tree up and down and that is the ultimate skill that makes them strong chess players even without any more modern software architecture.
steel-apron

this is an impossible task. grandmasters — from tarrasch to fischer — often make mistakes in their comments, analyzing irrelevant positions while ignoring the ones where real turning points occur. moreover, they don't really need their thought process to be accessible to the general public. fischer, for example, could have said much more, but he is extremely laconic. the computer lies. it only knows the number and cannot fundamentally provide a qualitative assessment of the position. so what is to be done? rely on your own talent, study the chess laws — those that both grandmasters with their unbridled imagination and computers are forced to obey. and one day, having played a large (or maybe not so large) number of games, you will suddenly realize that you no longer need other people's comments — and you can write your own laws on the chessboard. good luck!

p.s. grandmasters are not always wrong in their commentaries. often, between the lines, you can find seeds that, if they sprout, could bear good fruit — but fertile ground is required. and i repeat: these books are not written to multiply the number of competitors.

this is an impossible task. grandmasters — from tarrasch to fischer — often make mistakes in their comments, analyzing irrelevant positions while ignoring the ones where real turning points occur. moreover, they don't really need their thought process to be accessible to the general public. fischer, for example, could have said much more, but he is extremely laconic. the computer lies. it only knows the number and cannot fundamentally provide a qualitative assessment of the position. so what is to be done? rely on your own talent, study the chess laws — those that both grandmasters with their unbridled imagination and computers are forced to obey. and one day, having played a large (or maybe not so large) number of games, you will suddenly realize that you no longer need other people's comments — and you can write your own laws on the chessboard. good luck! p.s. grandmasters are not always wrong in their commentaries. often, between the lines, you can find seeds that, if they sprout, could bear good fruit — but fertile ground is required. and i repeat: these books are not written to multiply the number of competitors.
ChessPriyome

@midori-laser said ^

So I have a few questions:

When you’re analysing a game and get stuck trying to understand a move, where do you go for deeper insight? Do you consult stronger players, specific resources, or something else?

Stronger players, forums such as this. I personally would fire an engine on it if stuck and play out the position a few times to get some insight. Throw out the question to the internet in a forum and see what fish you catch. I will tell you most responses will probably be engine-verified because people dont like to be wrong in public, but its a nice experiment. Luckily, I've been playing chess since 1970 and being wrong does not bother me at all. It's why I'm not a liberal...but thats another conversation.

Local chess clubs: Set the board up and start walking through the game off a scoresheet. you will attract opinions because as chess players, we cannot help ourselves. When I started playing in the Fischer years (1970) the local library had a chess club in the basement on Friday nights with some strong players (a few became GM's) who would not be averse to pushing wood and telling you why it was a bad move, egos aside.

Discord: chess is alive and well on discord chats.

Find a friend: Studying chess with a friend is easily the best thing you can do. You bounce ideas off each other, get immediate feedback, etc.

With the growing number of AI tools available, is there one you’d recommend as being particularly strong for explaining chess ideas and decisions?

So, I use Copilot only because I only do chess on my personal comp. For work, we are married to Claude/Claude Code (software engineer).

I know that with extensive experimentation with copilot over several months you need to teach it to evaluate positions, make permanent rules, make it make sure squares are the correct color, pieces in the right place, knights can be a PITA to it. Often it misrepresents FEN diagrams. I make it confirm all square colors and piece placements based on Permanent rules. Even ASCII Boards are misrepresented. Frigging frustrating. I literally needed to make a rule that a knight controls the opposite color squares it is sitting on. It was too 'stupid' to understand it. It is literally like guiding a child sometimes.

I’ve had a Chess.com subscription for nearly a year, but I’ve found the instructional content a bit underwhelming. While the puzzles are fine, there seems to be a surprising lack of reading material. I tend to enjoy learning more through reading than through videos. Are there any resources you’d recommend that fit that preference?

The two you got are Good books. Absorb them completely.

BUT, you need to play, and play longer time controls and analyze your games afterwards, if you want to improve.

However, The Best books for YOU right now at your level (1300'ish):
Logical Chess: Move by Move — Irving Chernev <== You got this already!
Understanding Chess Move by Move — John Nunn <== next purchase
The Amateur’s Mind — Jeremy Silman <== Silman is in your wheelhouse
Simple Chess — Michael Stean <== A little heavier lift

My brain dump blog for chess:
https://chess-training.blogspot.com/

Addtionally, i have purchased Fishers' 60 Memorable Games and Cernev's Logical Chess. Chernev's book has been insightful and i would be keen for any recommendations of similar material

Given all that, I’d really appreciate any advice on how to better understand the reasoning behind moves during analysis.

Good luck. chess is an amazing game.

@midori-laser said [^](/forum/redirect/post/yO8hizvt) > So I have a few questions: > > When you’re analysing a game and get stuck trying to understand a move, where do you go for deeper insight? Do you consult stronger players, specific resources, or something else? > **Stronger players, forums such as this.** I personally would fire an engine on it if stuck and play out the position a few times to get some insight. Throw out the question to the internet in a forum and see what fish you catch. I will tell you most responses will probably be engine-verified because people dont like to be wrong in public, but its a nice experiment. Luckily, I've been playing chess since 1970 and being wrong does not bother me at all. It's why I'm not a liberal...but thats another conversation. **Local chess clubs:** Set the board up and start walking through the game off a scoresheet. you will attract opinions because as chess players, we cannot help ourselves. When I started playing in the Fischer years (1970) the local library had a chess club in the basement on Friday nights with some strong players (a few became GM's) who would not be averse to pushing wood and telling you why it was a bad move, egos aside. **Discord:** chess is alive and well on discord chats. **Find a friend:** Studying chess with a friend is easily the best thing you can do. You bounce ideas off each other, get immediate feedback, etc. > With the growing number of AI tools available, is there one you’d recommend as being particularly strong for explaining chess ideas and decisions? > So, I use **Copilot** only because I only do chess on my personal comp. For work, we are married to Claude/Claude Code (software engineer). I know that with extensive experimentation with copilot over several months you need to teach it to evaluate positions, make permanent rules, make it make sure squares are the correct color, pieces in the right place, knights can be a PITA to it. Often it misrepresents FEN diagrams. I make it confirm all square colors and piece placements based on Permanent rules. Even ASCII Boards are misrepresented. Frigging frustrating. I literally needed to make a rule that a knight controls the opposite color squares it is sitting on. It was too 'stupid' to understand it. It is literally like guiding a child sometimes. > I’ve had a Chess.com subscription for nearly a year, but I’ve found the instructional content a bit underwhelming. While the puzzles are fine, there seems to be a surprising lack of reading material. I tend to enjoy learning more through reading than through videos. Are there any resources you’d recommend that fit that preference? > The two you got are Good books. Absorb them completely. BUT, you need to play, and play longer time controls and analyze your games afterwards, if you want to improve. However, The Best books for YOU right now at your level (1300'ish): Logical Chess: Move by Move — Irving Chernev <== You got this already! Understanding Chess Move by Move — John Nunn <== next purchase The Amateur’s Mind — Jeremy Silman <== Silman is in your wheelhouse Simple Chess — Michael Stean <== A little heavier lift My brain dump blog for chess: https://chess-training.blogspot.com/ > Addtionally, i have purchased Fishers' 60 Memorable Games and Cernev's Logical Chess. Chernev's book has been insightful and i would be keen for any recommendations of similar material > > Given all that, I’d really appreciate any advice on how to better understand the reasoning behind moves during analysis. Good luck. chess is an amazing game.
kindaspongey

@midori-laser said in #1:

... I’ve started analysing games, both my own and those of top players. In doing so, I often find myself wondering why a particular move was played, especially at the highest level. ... When you’re analysing a game and get stuck trying to understand a move, where do you go for deeper insight? Do you consult stronger players, specific resources, or something else? ..., I’d really appreciate any advice on how to better understand the reasoning behind moves during analysis.

It may be that it just isn't worth the time to try to find out, but sometimes it just takes a little exploration. I remember looking at the position after 1 d4 e5 2 dxe5 Nc6 3 Nf3 Qe7 4 Bf4 Qb4+ 5 Bd2 Qxb2 6 Nc3 Nb4 and wondering why the standard reaction is 7 Nd4. I was wondering what would be wrong with 7 Rb1, so I just tried it and I was like: ohhh!
One option is to post a question, but, if you do that, be sure to be very clear about the specific position, the move number, and so on.

@midori-laser said in #1: > ... I’ve started analysing games, both my own and those of top players. In doing so, I often find myself wondering why a particular move was played, especially at the highest level. ... When you’re analysing a game and get stuck trying to understand a move, where do you go for deeper insight? Do you consult stronger players, specific resources, or something else? ..., I’d really appreciate any advice on how to better understand the reasoning behind moves during analysis. It may be that it just isn't worth the time to try to find out, but sometimes it just takes a little exploration. I remember looking at the position after 1 d4 e5 2 dxe5 Nc6 3 Nf3 Qe7 4 Bf4 Qb4+ 5 Bd2 Qxb2 6 Nc3 Nb4 and wondering why the standard reaction is 7 Nd4. I was wondering what would be wrong with 7 Rb1, so I just tried it and I was like: ohhh! One option is to post a question, but, if you do that, be sure to be very clear about the specific position, the move number, and so on.

This topic has been archived and can no longer be replied to.