Introduction
Have you ever wondered why can’t humans beat AI in chess? Well don’t worry, you’re not the only one to ask that question. In fact, this is one of the most popular questions that we get here.
So why can’t we humans beat computers? Are we too dumb? The answer is a bit more complicated than that but in essence, yeah, we are not that smart. Or at least we can’t calculate as well as a computer can so we can never reach the level these machines can.
It’s not a terminator scenario though, chess competitions are still going strong. And while cheating might be a legitimate concern, in reality most people rarely play with a cheater, check your games with an engine after your losses and you will see multiple blunders from your opponent every single time too!
What is Computer Chess
Computer Chess is a term which includes both hardware and software that can play a game of chess. This provides a great chance for people to play in the absence of human opponents, it’s also an amazing way to train yourself and others in making better moves thanks to the way computers play.
As early as the first smartphones, computer chess applications have become widely available to anyone in both free and paid versions, so it’s now possible for everyone to have a potential superhuman chess opponent in the palm of their hand at all times.
There are many different techniques and strategies employed to make a computer play chess, they go from simple algorithms to complex heuristics methods that build on and evaluate massive trees that represent different sequences of play. This enables computers to leverage their amazing calculation power to process millions of these moves at a time.
Computer chess is now considered a solved paradigm, and playing chess against a computer is now as easy as booting any computer, table or phone and using one of the many available chess engines.
History of Computer Chess
Pre-Computer Age
Before computers even existed the idea of a machine that could play chess was an intriguing concept that many inventors tried.
It wasn’t until 1770 that Hungarian inventor Farkas Kempelen became famous for creating the first machine that could play chess. It was a sensation back then and toured many countries until a fire destroyed the machine in 1854. Afterwards, the son of the machine’s last owner published the secrets of the machine and revealed that it was all an elaborate hoax, a trick. The machine was operated by a human inside the machine so it was a fraud or a mechanical illusion.
In 1912 Spanish engineer Leonardo Torres Quevedo created the first real machine able to play chess with a human for real. Called “La Ajedrecista” this machine played a king and rook endgame versus a knight. It couldn’t play a full game of chess and could only play with these pieces.
Even with these limitations this was an amazing achievement for the time. The machine was very popular back then and we could consider it the first computer game ever created in history.
After these two remarkable machines no one really tried to build anything else until computers started appearing for the first time in 1950.
Early Computer Age
It wasn’t long before chess enthusiasts tried to use computers to play chess. Mikhail Botvinnik was the first notable player who really dedicated himself to creating and researching a chess playing computer, he wrote several books on this subject. He held a degree in electrical engineering which helped his efforts, together with his reputation as a chess world champion.
The problem for Botvinnik was that he had to work with ancient computer hardware that was about 10 years behind the newest computers in the world at the time. The computers in the Soviet Union could hardly look only three moves in, which limited the ability of these machines to play properly.
Still his work at this time was very influential to many in the world of early computer chess software, with many of his ideas being used in subsequent efforts in the future.
Chess Playing Software
In the 70s the first fully fledged chess software started appearing in the US. Computer chess championships started emerging all over the country and the race was on to create the strongest machine that could play chess.
It was at this time with the limited computing capacity of the time that most of the modern paradigms for chess software were set. Engineers at the time had to implement clean and fast algorithms that could calculate great moves with limited calculations.
By the late 70s some chess machines could play chess but they were not that strong and their moves didn’t make much sense. It wasn’t until the 80s with the discovery of the microchip that we saw computers able to calculate massive amounts of moves and so they became capable of brute forcing calculations.
By the 90s supercomputers could reliably compete against masters but the best players in the world were still human, this changed with the famous chess match of 1997: Garry Kasparov vs Deep Blue.
After the turn of the century, most computers could best most human players and only “super” grand masters could even compete. But by the late 00s no human could compete with a computer in chess. Humanity was finally bested thanks to the brute forcing approach.
Beyond Super-Human Chess
While computers could beat humans 100% of the time after these advancements there are still things to improve in chess software. This is why even today there are multiple efforts to create and improve engines like Stockfish, GNU Chess and Fruit. `
It’s in this context that AlphaZero was born, a computer program designed to play Shogi, Go and Chess. It uses an algorithm similar to AlphaGo Zero. This program doesn’t use opening books or endgame tables and instead uses neural networks to train itself to play chess.
This approach has led to an innovative way for machines to play chess, this approach is so powerful that this program is capable of playing other types of games that would seem impossible in the past, for example OpenAI used reinforcement learning to teach 5 bots how to play DOTA2 and in 2019 they beat the world champions OG in a real match.
Can a Human Beat a Computer in Chess
If you have been reading this article seriously then you already know the answer. Since the late 90s humans have not been able to beat computers in chess. In fact, only the best of the best could even hope to compete at the level these machines played even back then.
For a normal human, even the weakest CPU of a consumer-grade phone is enough to win every single game. A cheap $100 phone is more than strong enough to calculate millions of lines at a time which is simply impossible for humans.
This has many implications for chess but most people don’t see computers as opponents now, instead they use computer analysis to better their play with objective measurements of every move.
Previously you needed a master to evaluate your moves to see how to improve, but now anyone can boot up stockfish in their computer or phone and review their games easily. This has led to a slight increase in the level of play of humans.
Did Computers Ruin Chess
No, of course not. Just like calculators didn’t ruin math, computers are just a tool. But they certainly changed things in chess. Both in professional and amateur chess most people use a chess engine as a practice tool.
Most people understand the strengths of a computer and accept the fact that they can calculate so much more than us. Even Magnus Carlsen doesn’t play versus computers very often because it’s useless. No human can play like a computer ever.
Conclusion
In the future we can expect computer chess engines to continue improving alongside computers themselves. As of today, chess as a game is still not solved (unlike checkers) so computers still got a long way to go to do that.
In theory quantum computers could bring about another revolution for computer chess, but in that case, I think chess is going to be the least of our concerns. Instead, we should be aiming higher like true artificial intellect and how it could play chess by simulating how a human being thinks for example.