Is Ai the End of Chess?
For as long as chess has been around, it has been a way to match two minds against one another, competing to best the other through a combination of analytical skill and intuition. However, has chess become pointless through the development of increasingly advanced AI chess engines? Is AI the end of chess?
In short, the answer is both yes and no.
It is true that the best chess engines are already unbeatable by humans. Rapid developments are being made to AI chess engines. AlphaZero, currently the strongest engine, utilizes machine learning. AlphaZero is more efficient than Stockfish, the previous AI chess champion, in that AlphaZero searches fewer positions. Furthermore, AlphaZero beat Stockfish after only 4 hours of learning.
In recent years, due to the prevalence of such chess engines, the game has lost much of its uncertainty. In fact, among players rated 2750 and above, the draw rate has increased dramatically over the years. One can see how this increase may seem to be an indicator of chess becoming more and more “solved.” Some argue that the high percentage of draws in high rated chess detracts from the game’s entertainment.
However, at its core, chess is a game that tests the skill of humans. This means even if a game ends in a draw, it doesn’t mean that it was pointless or boring. Chess is about the process and the battle that takes place during the game, not the end result. Yes, it is entirely possible that in the future AI will have optimized every move from start to finish in a game, and then chess will be a matter of pure memorization.
But, before we reach that point, chess can be adapted in many different ways to bring back the aspects of intuition and powerful risks that make chess thrilling. One method is to play shorter games. This is already used to break ties, as seen in the 2018 World Championship. In order to break a tie of 12 draws, the players played rapid chess, with 25 minutes per player and a 10 second increment per move. This forces players to think faster on the fly, thus relying more on instinct and quick thinking and allowing more errors to capitalize on.
Alternatively, there are chess variants such as Fischer random, displacement chess, or pre-chess that can make the game fresher and more entertaining. Of course, one can make the argument that such variants are not actually chess, as the base rules are different, but that is a whole different argument altogether.
Something else to note is that these problems with Ai and draws and chess becoming more “solved” is only really an issue at very high levels of competitive play. The classification of “Master” according to the US Chess Federation (USCF) is 2200 ELO or above, which is a percentile of about 99.1. Furthermore, a “Senior Master”, someone who is 2400 ELO and above, would be placed at around the 99.7th percentile.
The vast majority of competitive tournament players are nowhere close to this level, not to mention casual players. As an intermediate player myself, I put a sizable number of years of my childhood into chess and only peaked at 1574. I can confidently say that to an average player, the rise in AI chess engines is something positive. These engines allow players to analyze theirs and other games in order to learn and improve.
Ultimately, AI may be the downfall of high-level chess, but in the near future, chess is and will continue to be the game we know and love.
Written by Bryan Xiao & Edited by Alexander Fleiss