Even if it may be complicated, is it possible with the present technology?
$\begingroup$ $\endgroup$
2 
- 6$\begingroup$ Chess is still not a solved game, therefore the answer is unknown. And the answer would be yes only if the optimal play in chess ends with a draw. $\endgroup$rus9384– rus93842024-07-23 21:42:27 +00:00Commented Jul 23, 2024 at 21:42
- 2$\begingroup$ @rus9384, I suggest writing that as an answer rather than a comment, so we can upvote it and accept it and the question can be treated as answered. The SE platform discourages writing answers in the comment field. No need to make it longer, it is fine if answers are short. $\endgroup$D.W.– D.W. ♦2024-07-23 21:52:45 +00:00Commented Jul 23, 2024 at 21:52
Add a comment |
2 Answers
$\begingroup$ $\endgroup$
Currently, the best chess AI engines are unbeatable for humans. However, they are not invincible, they are just better than any human. Also, if the level of humans would improve, we will have new training data to feed to the chess engines so that they will also improve. Probably, we will never beat chess engines again.
For having a comparison, citing Wikipedia: "Stockfish [...] as of July 2024, is the strongest CPU chess engine in the world with an estimated Elo rating of 3634", while the Elo of the best human chess player on Heart (Magnus Carlsen) on Chess.com is 2832. This is a very wide gap which makes the chess engine practically unbeatable.
However, as the comment of rus9384 states, chess game is not solved, so it is not currently possible to build a theoretically invincible chess engine. In general, it is an open problem.
$\begingroup$ $\endgroup$
In practice, no, as others already pointed out.
In theory, however, yes, and it is in fact quite easy. There's a very simple algorithm called 'minimax' that plays theoretically optimal. This means that at least as one of the two players ('black' or 'white') it is invincible.
The problem, however, is that on a real computer this algorithm would take an extremely long time to finish.
