Your network blocks the Lichess assets!

lichess.org
Donate

AI Chess Coaches?

This gives me the idea of minimaxing plans! It brings an engine closer to human thought. Basically use the planning DSL as the latent space. Instead of calculating the best move the opponent would make, the question would be what is the most likely plan the opponent is pursuing, and then what is the best answer to that plan. This might even make the need for exit conditions - which was kind of ugly to start with - unnecessary.

This gives me the idea of minimaxing plans! It brings an engine closer to human thought. Basically use the planning DSL as the latent space. Instead of calculating the best move the opponent would make, the question would be what is the most likely plan the opponent is pursuing, and then what is the best answer to that plan. This might even make the need for exit conditions - which was kind of ugly to start with - unnecessary.

@Noobmasterplayer123 said in #3:

Thanks for the shoutout to my ChessAgine project. AI coaching, especially with LLMs on their own, is very hard to almost impossible with current LLM tech. The thing is reddit companies and social media have made AI in chess so hard to work with; people automatically slap labels on newer projects that are trying to work from different angles, like my agine project. To this day, people confuse my project with an AI coach, or I hear things like "Oh its wrong haha your project is no way near a chess coach," but they don't check the agine docs and read that it's just an engine to the LLM interface, not meant to be a coach to start with lol. You will encounter this as you build, I do warn you. Also, my interface is a work in progress, as more devs get interested, we will have more sophisticated systems in future, but again, no way near a real human chess coach. I think the premise of "AI Chess coach" in 2026 is just a marketing scheme, and you are right, this will take a lot of work.

Just keep building

Good luck!

Yep, ChessAgine is a cool product, and I think it is nice to see where LLM's can take us, I do think the work you are doing will be the backbone or at least contribute to the idea of AI Chess Coaches in the future.

Keep progressing this buddy and keep working on it :)

@Noobmasterplayer123 said in #3: > Thanks for the shoutout to my ChessAgine project. AI coaching, especially with LLMs on their own, is very hard to almost impossible with current LLM tech. The thing is reddit companies and social media have made AI in chess so hard to work with; people automatically slap labels on newer projects that are trying to work from different angles, like my agine project. To this day, people confuse my project with an AI coach, or I hear things like "Oh its wrong haha your project is no way near a chess coach," but they don't check the agine docs and read that it's just an engine to the LLM interface, not meant to be a coach to start with lol. You will encounter this as you build, I do warn you. Also, my interface is a work in progress, as more devs get interested, we will have more sophisticated systems in future, but again, no way near a real human chess coach. I think the premise of "AI Chess coach" in 2026 is just a marketing scheme, and you are right, this will take a lot of work. > > Just keep building > > Good luck! Yep, ChessAgine is a cool product, and I think it is nice to see where LLM's can take us, I do think the work you are doing will be the backbone or at least contribute to the idea of AI Chess Coaches in the future. Keep progressing this buddy and keep working on it :)

@TotalNoob69 said in #9:

how do you define and codify a plan? Also, how can you heuristically determine what the plan is

That's the hard part :D I have no idea. I believe starting with generating plans would be easier. Then, for existing games, just generate more plans and calculate the likelihood for each that it was what was intended. Third step, apply it to women and get rich.

Yeah, I think this is why I am kinda drawn to the idea, because forget Chess for a moment, this has far wider implications, imagine in business, you had in the variables and it can generate a plan? Right now, AI gives you the cliff notes of what possible plans are there, but nothing you can implement with full confidence. As I work in the field, it is a fascinating thing to think about.

@TotalNoob69 said in #9: > > how do you define and codify a plan? Also, how can you heuristically determine what the plan is > > That's the hard part :D I have no idea. I believe starting with generating plans would be easier. Then, for existing games, just generate more plans and calculate the likelihood for each that it was what was intended. Third step, apply it to women and get rich. Yeah, I think this is why I am kinda drawn to the idea, because forget Chess for a moment, this has far wider implications, imagine in business, you had in the variables and it can generate a plan? Right now, AI gives you the cliff notes of what possible plans are there, but nothing you can implement with full confidence. As I work in the field, it is a fascinating thing to think about.

@TotalNoob69 said in #11:

This gives me the idea of minimaxing plans! It brings an engine closer to human thought. Basically use the planning DSL as the latent space. Instead of calculating the best move the opponent would make, the question would be what is the most likely plan the opponent is pursuing, and then what is the best answer to that plan. This might even make the need for exit conditions - which was kind of ugly to start with - unnecessary.

Funny enough, that gets us back to LLMs, sort of. Instead of "what is the best move in the position" you would ask "what is the best move given the preceding moves and the style of the players, which are also determined by preceding moves from other games".

@TotalNoob69 said in #11: > This gives me the idea of minimaxing plans! It brings an engine closer to human thought. Basically use the planning DSL as the latent space. Instead of calculating the best move the opponent would make, the question would be what is the most likely plan the opponent is pursuing, and then what is the best answer to that plan. This might even make the need for exit conditions - which was kind of ugly to start with - unnecessary. Funny enough, that gets us back to LLMs, sort of. Instead of "what is the best move in the position" you would ask "what is the best move given the preceding moves and the style of the players, which are also determined by preceding moves from other games".

@TotalNoob69 said in #11:

This gives me the idea of minimaxing plans! It brings an engine closer to human thought. Basically use the planning DSL as the latent space. Instead of calculating the best move the opponent would make, the question would be what is the most likely plan the opponent is pursuing, and then what is the best answer to that plan. This might even make the need for exit conditions - which was kind of ugly to start with - unnecessary.

This is interesting, I am going to have to let this stew a bit. Karpovian style play against your opponents likely plans?

@TotalNoob69 said in #11: > This gives me the idea of minimaxing plans! It brings an engine closer to human thought. Basically use the planning DSL as the latent space. Instead of calculating the best move the opponent would make, the question would be what is the most likely plan the opponent is pursuing, and then what is the best answer to that plan. This might even make the need for exit conditions - which was kind of ugly to start with - unnecessary. This is interesting, I am going to have to let this stew a bit. Karpovian style play against your opponents likely plans?

@HollowLeaf said in #15:

This is interesting, I am going to have to let this stew a bit. Karpovian style play against your opponents likely plans?

I am thinking this brings back the "contempt" concept, which would not really be contempt, but calculating the lines the opponent is likely to make based on experience and previous moves, rather than expecting them to make the SF moves. This enables engines that gambit.

@HollowLeaf said in #15: > This is interesting, I am going to have to let this stew a bit. Karpovian style play against your opponents likely plans? I am thinking this brings back the "contempt" concept, which would not really be contempt, but calculating the lines the opponent is likely to make based on experience and previous moves, rather than expecting them to make the SF moves. This enables engines that gambit.

@TotalNoob69 said in #14:

Funny enough, that gets us back to LLMs, sort of. Instead of "what is the best move in the position" you would ask "what is the best move given the preceding moves and the style of the players, which are also determined by preceding moves from other games".

Oh amazing thought!!!! Oh that is pretty cool as an idea. I have been updating my opponent prep tool, and I think that actually adds a great new angle!!!! Right now, I am taking your games and the opponents games and comparing openings, but it would be awesome to compare styles, and find openings that contradicts how they like to play.

i.e. Your opponent is a sharp attacking player, so therefore, a possible way of playing will be to not allow them to play their style of play, take them out of their comfort zone. Now that is an interesting thought for the day and will cost me hours of sleep...

@TotalNoob69 said in #14: > Funny enough, that gets us back to LLMs, sort of. Instead of "what is the best move in the position" you would ask "what is the best move given the preceding moves and the style of the players, which are also determined by preceding moves from other games". Oh amazing thought!!!! Oh that is pretty cool as an idea. I have been updating my opponent prep tool, and I think that actually adds a great new angle!!!! Right now, I am taking your games and the opponents games and comparing openings, but it would be awesome to compare styles, and find openings that contradicts how they like to play. i.e. Your opponent is a sharp attacking player, so therefore, a possible way of playing will be to not allow them to play their style of play, take them out of their comfort zone. Now that is an interesting thought for the day and will cost me hours of sleep...

Very meaningful by totalnoob69

Very meaningful by totalnoob69

@Sahuadvik said in #18:

Very meaningful by totalnoob69

Yep, it has given me tons to think about.

@Sahuadvik said in #18: > Very meaningful by totalnoob69 Yep, it has given me tons to think about.

I guess I'll share my opinion although it won't make me any friends.

The ChessMaster series can analyze games and construct sentences and even apply a human voice reading those sentences, and it's pretty good.

Previously, there was Kasparov's Gambit (1993) which offered live commentary on your current game; then Fritz 4 which had a "coach" mode kibitzing to let you know as you blunder, and recommending takebacks... trying to get you to focus on the mistakes which matter most.

Later there was Maurice Ashley Teaches Chess, a powerful teaching tool explaining fundamentals and providing wonderful exercises; and Bobby Fischer Teaches Chess (book and software) around the same timeframe with interactive, customizable tutorials.

Since the 90's there have been no major advances in "AI coaching" unless you count what ChessMaster does. Sure, databases and endgame tablebases are great, but everything's the same old technology with new marketing. But why?

  1. Old tools are good enough for teaching most players (most players are beginner to intermediate players), so there's not much R&D money in AI teaching to begin with.
  2. Most players need to work on openings, tactics, and endgames but you can't make money teaching endgames or tactics.
  3. It's not as if someone with poor tactics or endgames can pull off a positional sacrifice like a master can.
I guess I'll share my opinion although it won't make me any friends. The ChessMaster series can analyze games and construct sentences and even apply a human voice reading those sentences, and it's pretty good. Previously, there was Kasparov's Gambit (1993) which offered live commentary on your current game; then Fritz 4 which had a "coach" mode kibitzing to let you know as you blunder, and recommending takebacks... trying to get you to focus on the mistakes which matter most. Later there was Maurice Ashley Teaches Chess, a powerful teaching tool explaining fundamentals and providing wonderful exercises; and Bobby Fischer Teaches Chess (book and software) around the same timeframe with interactive, customizable tutorials. Since the 90's there have been no major advances in "AI coaching" unless you count what ChessMaster does. Sure, databases and endgame tablebases are great, but everything's the same old technology with new marketing. But why? 1. Old tools are good enough for teaching most players (most players are beginner to intermediate players), so there's not much R&D money in AI teaching to begin with. 2. Most players need to work on openings, tactics, and endgames but you can't make money teaching endgames or tactics. 3. It's not as if someone with poor tactics or endgames can pull off a positional sacrifice like a master can.