Personalized AI learning tools • page 1/1 • General Chess Discussion • lichess.org

mckinsey_and_co

What are the best / your favourite tailored (to your game history / style) AI learning tools you have found that generate tailored learning programs?

T_deus_ex_T

I don't use AI... For training I play against StockFish with various piece odds, this has helped me tremendously.

CamilleQueenOfTheBot

AI is not so good at chess and for something as specific as doing a training program I would not trust it at a high level.

Unless you want to stay on very basic stuff, like we do on Camille, the chess engine analyze your game , the AI analyze the results and propose exercise based on frequent mistake.

AI is not so good at chess and for something as specific as doing a training program I would not trust it at a high level. Unless you want to stay on very basic stuff, like we do on Camille, the chess engine analyze your game , the AI analyze the results and propose exercise based on frequent mistake.

ChessPriyome

AI is good for coming up with study plans for a book you want to read.
AI is good for collecting thematic chapters across multiple books to concentrate on a single aspect of chess. For instance, "IQP".
Engines are good at analysis - AI, not so much.

I did have some success pushing rules onto it so it understands positions better, but it 'forgets' where pieces are, what color squares they are on, etc. I'm going to work with CoPilot and Claude in putting together a set of skills it can apply to chess so it avoids dumb evals.

There are things it does well and things not so well, but trial and error have to be tested.

ChessPriyome/ChessDBA, Oracle Professional

AI is good for coming up with study plans for a book you want to read. AI is good for collecting thematic chapters across multiple books to concentrate on a single aspect of chess. For instance, "IQP". Engines are good at analysis - AI, not so much. I did have some success pushing rules onto it so it understands positions better, but it 'forgets' where pieces are, what color squares they are on, etc. I'm going to work with CoPilot and Claude in putting together a set of skills it can apply to chess so it avoids dumb evals. There are things it does well and things not so well, but trial and error have to be tested. ChessPriyome/ChessDBA, Oracle Professional

CamilleQueenOfTheBot

@ChessPriyome said ^

AI is good for coming up with study plans for a book you want to read.
AI is good for collecting thematic chapters across multiple books to concentrate on a single aspect of chess. For instance, "IQP".
Engines are good at analysis - AI, not so much.

I did have some success pushing rules onto it so it understands positions better, but it 'forgets' where pieces are, what color squares they are on,...

Yes the idea is :
use none AI tool to gather data and analysis this data, use the AI for putting word on the results and organize them on section. Working like this will work.

Once I had opus 5 do some direct analysis of results , the results were a perfect 50/50 when one bot did not win one match. Opus forgot that has each bot were switching side .. It was plainly written in the table of results..

@ChessPriyome said [^](/forum/redirect/post/VlOYv4Dx) > AI is good for coming up with study plans for a book you want to read. > AI is good for collecting thematic chapters across multiple books to concentrate on a single aspect of chess. For instance, "IQP". > Engines are good at analysis - AI, not so much. > > I did have some success pushing rules onto it so it understands positions better, but it 'forgets' where pieces are, what color squares they are on,... Yes the idea is : use none AI tool to gather data and analysis this data, use the AI for putting word on the results and organize them on section. Working like this will work. Once I had opus 5 do some direct analysis of results , the results were a perfect 50/50 when one bot did not win one match. Opus forgot that has each bot were switching side .. It was plainly written in the table of results..