Listen to this story
In 2021, Nikhil Kamath, founder of Zerodha, defeated five-time world champion Vishwanathan Anand in chess with the help of computers (he confessed later on) at a celebrity fundraiser. The controversy sparked discussions around the use of AI in the game of chess.
As India is all set to host the 44th edition of the Chess Olympiad in Mahabalipuram starting on July 28, let’s look at how AI has impacted the game of chess.
Sign up for your weekly dose of what's up in emerging technology.
AI in chess
The earliest mention of technology in chess can be traced back to the 18th century when Austrian empress Maria Theresa commissioned a chess-playing machine. Many players competed against the ‘Mechanical Turk’, thinking it was an automated machine. However, it turned out to be a scam. A human hidden inside the machine was operating it.
In the mid-1940s, British mathematician Alan Turing began theorising how a computer could play chess against a human. In 1949, Claude Shannon published a seminal paper describing a potential program to do exactly that. In 1950, Alan Turing created a program capable of playing chess. Soon after, the Dietrich Prinz and Bernstein chess program burst into the scene.
Computer chess appeared for the first time in the 1970s. MicroChess, the first commercial chess program for microcomputers, in 1976; Chess Challenger in 1977; and Sargon, which won the world’s first computer chess tournament for microcomputers, in 1978.
The robotic chess computers came about in the 1980s. Boris Handroid, Novag Robot Adversary and Milton Bradley Grandmaster are some examples. The most popular was Chessmaster 2000, which ruled the chess video and computer games industry for the next two decades.
As chess computers were gaining popularity in the 1980s, Gary Kasparov, the then world chess champion, claimed AI-driven chess engines could not defeat top-level chess grandmasters. However, in 1989 and 1996, Kasparov beat IBM’s powerful chess engines, Deep Thought and Deep Blue.
Things started to change in the late 1990s. In 1997, Deep Blue defeated Kasparov. A year later, Kasparov came up with the idea of Cyborg chess or centaur chess, in which human and computer skills are combined to up the level of the game. The first cyborg chess was held in 1998.
In 2017, AlphaZero, a computer program developed by DeepMind, defeated the world’s strongest chess engine Stockfish. AlphaZero used the reinforcement learning technique in which the algorithm mimicked humans’ learning process to train its neural networks.
In 2018, TalkChess.com released Leela Chess Zero, developed by Gary Linscott (who also developed Stockfish). Without having any chess-specific knowledge, Leela Chess Zero learned the game based on deep reinforcement learning using an open-source implementation of AlphaZero.
In 2019, DeepMind came up with another algorithm based on reinforcement learning called MuZero.
Chess players use AI-driven chess engines to analyse their and competitors’ games. As a result, AI has helped in improving the quality of games.
Post pandemic a lot of chess competitions were moved online. In the European Online Chess Championship, as many as 80 participants were disqualified for cheating. FIDE, the international chess body, has approved an artificial intelligence-driven behaviour-tracking module for the FIDE Online Arena games. Chess.com, an internet chess server, uses a cheat detection system to assess the probability of a human player matching the moves of a chess engine or surpassing the games of some of the greatest chess players with the help of a statistical model. DeepMind is also working to develop a new cheat detection software.
AI has also brought down the cost and effort of training and helped develop new chess strategies.
AI has indeed changed the dynamics of the game. However, using AI in chess has raised a few issues. Computer chess engines have significantly improved gameplay. However, people have also raised concerns that players of this age depend too much on machine-driven analysis.
Even when it comes to detecting cheating, AI poses a few issues. First, there is a possibility a player might be wrongly red-flagged by AI. For example, a Chess.com player and grandmaster, Akshat Chandra, was banned after a win against Hikaru as his moves supposedly matched Komodo, a strong positional chess engine. Though Chandra has been proved innocent, his reputation took a hit.
Chess engines and deep learning-based neural networks present enormous possibilities. Moreover, the complex nature and the strategic orientation of the game have provided a ground for assessing any progress in the field of artificial intelligence. “They (games) are the perfect platform to develop and test ideas for AI algorithms. It’s very efficient to use games for AI development, as you can run thousands of experiments in parallel on computers in the cloud and often faster than real-time, and generate as much training data as your systems need to learn from. Conveniently, games also normally have a clear objective or score, so it is easy to measure the progress of the algorithms to see if they are incrementally improving over time, and therefore if the research is going in the right direction,” said DeepMind cofounder Demis Hassabis.