lichess.org
Donate

Is Stockfish an AI Engine

Hello Everyone ,
My Question which I cannot understand is about the powerful engine Stockfish .

Is Stockfish AI (Artificial Intelligence) Engine like Fat Fritz and Leela Chess Zero ? If so which version of Stockfish is AI .

Is Stockfish NNUE better than Stockfish 12 or is Stockfish 12 better than Stockfish NNUE ?

Please help me by telling the answer

Hello Everyone , My Question which I cannot understand is about the powerful engine Stockfish . Is Stockfish AI (Artificial Intelligence) Engine like Fat Fritz and Leela Chess Zero ? If so which version of Stockfish is AI . Is Stockfish NNUE better than Stockfish 12 or is Stockfish 12 better than Stockfish NNUE ? Please help me by telling the answer

Depends on your definition of AI. But basically any software that can play chess is AI.

Stockfish prior to last one is not machine learning or neural net brand of AI. Stockfish NNUE is combination of Neural net evaluation and traditional alpha-beta tree search. While leela Chess 0 and a0 use monte-carlo tree search (without the random part like the original monte carlo searc did). Which will make SF tactically stronger than leela and a0.

Depends on your definition of AI. But basically any software that can play chess is AI. Stockfish prior to last one is not machine learning or neural net brand of AI. Stockfish NNUE is combination of Neural net evaluation and traditional alpha-beta tree search. While leela Chess 0 and a0 use monte-carlo tree search (without the random part like the original monte carlo searc did). Which will make SF tactically stronger than leela and a0.

Of course it is Sargon Chess running on commadore 64 was AI. AI is not something new. There just are some developments that have made something more efficient. Alpha-beta search (i.e if you find refutation to your move that loses a pawn there is no need to see if there opponents move that wins even more) is simulation of how people do analysis on situation. Evaluation function tries to codify human knowledge like

  • king not directly threated
  • Knights are worth in the centre of the board
    That is very definition of AI emulating human decission making. just that success has been rather modest outside board games.

Not all AI is machine learning. A0 and SF 14 evaluation are. Just that aout 20 years ago would havebeen hugely expensive effort to train such neural net. it aint something new that neural nets can learn game position evaluation. Backgammon that happened in late eighties if I remember correctly. Then it needed main frame computer. Late nineties it ran on normal laptop. Chess eval training would have economicallly unfeasible

Of course it is Sargon Chess running on commadore 64 was AI. AI is not something new. There just are some developments that have made something more efficient. Alpha-beta search (i.e if you find refutation to your move that loses a pawn there is no need to see if there opponents move that wins even more) is simulation of how people do analysis on situation. Evaluation function tries to codify human knowledge like - king not directly threated - Knights are worth in the centre of the board That is very definition of AI emulating human decission making. just that success has been rather modest outside board games. Not all AI is machine learning. A0 and SF 14 evaluation are. Just that aout 20 years ago would havebeen hugely expensive effort to train such neural net. it aint something new that neural nets can learn game position evaluation. Backgammon that happened in late eighties if I remember correctly. Then it needed main frame computer. Late nineties it ran on normal laptop. Chess eval training would have economicallly unfeasible

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