From Roberta Duarte
If the series “Gambito da Rainha” were to play today, the protagonist would be replaced by artificial intelligence
In the service of British intelligence, mathematician Alan Turing broke the Nazi Enigma code, which was crucial to the Allied victory in World War II (his story is told in the film “The Game of Imitation”, but you should already know). . Considered a hero, he focused his work in the field of computer science and artificial intelligence (AI), where he pioneered. It was his idea to set up a computer for human activity.
In 1948, Turing began working with his colleague David Champernowne on an algorithm that played chess. The Turochamp algorithm – the combination of the names Turing and Champernowne – was completed in 1950, but due to computing time constraints, the code could not be implemented. It was left to Turing and his other colleague, also the computer scientist Alick Glennie, to trace the algorithm in addition to logic, of course, with pencil and paper. In the same year the mathematician published his famous article “Computing Machinery and Intelligence”.
“Can machines think?” This is how the paper began in which he presented the Turing test: two people and a machine confront each other in a game with questions and answers. If the machine succeeds in tricking a player into thinking he is human, it has passed the test. The foundation was born that gave rise to the idea of artificial intelligence as we know it today.
However, Turing died in 1954 before seeing his code running on a computer. But his work opened many doors and the concept of a chess game machine remained a challenge for future scientists who developed and studied the subject for decades.
Cut to 1996 when things get even more interesting. That year, IBM introduced DeepBlue, a computer that calculates positions for playing chess. The world chess champion was the Russian Garry Kasparov and he was asked to play against the machine. On the one hand Kasparov, who represents humanity, on the other hand DeepBlue in the name of artificial intelligence. The Russian emerged victorious but warned that he would likely be the last person to win the belt against a computer.
Said and done. The following year he lost to an updated version of DeepBlue. Our team representative did not digest the defeat well and accused IBM of fraud. Years later, he admitted he hadn’t handled the situation well and even wrote a book on artificial intelligence in 2017.
Does the story end there? That was just the beginning. Since then, more and more algorithms have been released that use artificial intelligence to play chess. In 2010 the Top Chess Engine Championship (TCEC) started, a competition between computers, the aim of which is to find the best chess algorithm. The players are invited by the organization of the event, which lasts a few months. The Stockfish model that keeps this name has been crowned champion ten times.
In 2017, DeepMind, an AI-focused company, introduced AlphaZero, an artificial intelligence that can play chess, go, and shogi. The computer received the basic rules of chess and learned the game on its own. He played against himself countless times and in four hours the algorithm was a specialist in chess.
AlphaZero played a game against the Stockfish computer, winner of this year’s Top Chess Engine Championship. He managed to defeat the champion and become an honorary holder. If a sophisticated computer couldn’t beat it, what about a human?
We don’t know what the future will look like, but we can throw in the towel at chess. Kasparov was right: he was the last champion mankind had.
Roberta Duarte is a physicist with a PhD in astrophysics from the University of São Paulo (USP) and works with applications of artificial intelligence in black hole astrophysics.
Sign up for the Serrapilheira newsletter for more news from the institute and the Ciência Fundamental blog. Do you have an agenda? Here’s how you can work together.