4 ms·

Can you give an example? I haven't seen any that are indistinguishable from humans

anbende 6y ago
Well, apparently Chess24 has an engine that will let you play against a simulated Magnus Carlsen at various stages and skill-levels of his career. So you can set some engines not to pick perfect moves but to pick moves consistent with a particular rating or style...
- reidmcy 6y ago
  I don't think Play Magnus is close to indistinguishable, it's a modified version of Stockfish on the backend, having the same rating doesn't mean you have the same style
  - tutfbhuf 6y ago
    Yup. It's absolutely possible to determine whether you play versus a strength reduced/weak engine or a human. Humans make human plans and play according to human best practices, which you can find in chess literature. Engines, however, do not make long term plans and often play very inhuman (perfectly precise, unobvious tactical) moves, that often suprises even on grandmaster level.
    - reidmcy 6y ago
      I'm not sure if saying engines don't make long term plans is accurate, since we can set them to plan an arbitrary number of moves ahead. I think you mean that they don't have a theory of mind, and thus don't modulate their play based on their opponents.
      - watwut 6y ago
        No, he means abstract plans. Abstract plan is not exact sequence of moves, but rather general idea of where you want to be in far future. It could be modulated based on opponent, but does not have to be.
sudosysgen 6y ago
HIARCS or Lucas Chess are engines that have been made explicitly to play like a human. You won't be able to distinguish them, with some human play in conjunction, from a very strong human player.
- reidmcy 6y ago
  Thanks, those do look like they are attempting to play like humans, but neither says how that claim was tested. How would you test the their ability to imitate humans?
  - sudosysgen 6y ago
    By cheating using them and not getting called out anymore than a human player of the same elo.
    - reidmcy 6y ago
      That doesn't help you prove they're indisputable from humans, even if you admitted to using them to cheat. You said some engines are indisputable from humans, can you give me some evidence of this?
pashamur 6y ago
AlphaZero / LeelaChess - neural net / genetic learning based ones; although they're not really commercially available, technically anyone can download the code to LeelaChess and train it if given enough CPU power. They learn chess by repeatedly playing chess against itself and in such a way simulate the organic development that has occurred in human play as well. That's the next generation of chess engines - AlphaZero trounced Stockfish in a match of 100 games with 28 wins and 72 draws and not a single loss. And if you decrease the power a little bit (take an earlier generation), it makes significantly more human-like moves than Stockfish.

Well, apparently Chess24 has an engine that will let you play against a simulated Magnus Carlsen at various stages and skill-levels of his career. So you can set some engines not to pick perfect moves but to pick moves consistent with a particular rating or style...

reidmcy 6y ago
I don't think Play Magnus is close to indistinguishable, it's a modified version of Stockfish on the backend, having the same rating doesn't mean you have the same style
- tutfbhuf 6y ago
  Yup. It's absolutely possible to determine whether you play versus a strength reduced/weak engine or a human. Humans make human plans and play according to human best practices, which you can find in chess literature. Engines, however, do not make long term plans and often play very inhuman (perfectly precise, unobvious tactical) moves, that often suprises even on grandmaster level.
  - reidmcy 6y ago
    I'm not sure if saying engines don't make long term plans is accurate, since we can set them to plan an arbitrary number of moves ahead. I think you mean that they don't have a theory of mind, and thus don't modulate their play based on their opponents.
    - watwut 6y ago
      No, he means abstract plans. Abstract plan is not exact sequence of moves, but rather general idea of where you want to be in far future. It could be modulated based on opponent, but does not have to be.

sudosysgen 6y ago

HIARCS or Lucas Chess are engines that have been made explicitly to play like a human. You won't be able to distinguish them, with some human play in conjunction, from a very strong human player.

reidmcy 6y ago
Thanks, those do look like they are attempting to play like humans, but neither says how that claim was tested. How would you test the their ability to imitate humans?
- sudosysgen 6y ago
  By cheating using them and not getting called out anymore than a human player of the same elo.
  - reidmcy 6y ago
    That doesn't help you prove they're indisputable from humans, even if you admitted to using them to cheat. You said some engines are indisputable from humans, can you give me some evidence of this?

pashamur 6y ago

AlphaZero / LeelaChess - neural net / genetic learning based ones; although they're not really commercially available, technically anyone can download the code to LeelaChess and train it if given enough CPU power. They learn chess by repeatedly playing chess against itself and in such a way simulate the organic development that has occurred in human play as well. That's the next generation of chess engines - AlphaZero trounced Stockfish in a match of 100 games with 28 wins and 72 draws and not a single loss. And if you decrease the power a little bit (take an earlier generation), it makes significantly more human-like moves than Stockfish.