Cheating problem in Go

The story was reported in multiple news outlets, including CNN, where we first learned about the scandal. The Chinese Weiqi (Go) Association spent weeks investigating the matter, but found no evidence of cheating. However, the scandal does raise questions about the future of the game, with many parallels to the problem that chess faces.

You can watch the entire annotated game – if you have five hours for it

Jiuheng He, who researches AI at Cornell University and is an avid Go player, said to CNN: “Human experts used to dominate the whole realm. Now we have to accept a non-human actor who has expertise, maybe even has exceeded the human experts. So how are we going to deal with it?” 

The game of Go is played on a 19 x 19 board and has only one piece type (similar to checkers) and simple rules. But it is profoundly complex. There are an astonishing 10¹⁷⁰ possible board configurations, making the game immeasurably more complex than chess. Go was completely out of the scope of computer intelligence.

But then Google’s AI lab, DeepMind, developed AlphaGo, which was trained on thousands of human amateur and professional games. In 2016 it beat the 18-time world champion Lee Se-dol of South Korea in a widely-publicized match. Lee announced his retirement three years later, citing the match as the reason. The next iteration, AlphaGo Zero, learnt by simply playing games against itself. In doing so, it quickly surpassed human level of play and defeated the first AlphaGo version by 100 games to zero.

Since 2016 there have been a number of Go cheating scandals. In 2020 a South Korean court sentenced two people to a year in prison after they were caught using AI in an official competition, using a camera and earphones and receiving computer moves from an accomplice outside. Players fears the sport will lose its relevance if Go organizations cannot guarantee clean games. The situation is especially precarious in online Go leagues, where it is estimated that top players frequently use AI to enhance their game.

Players generally also use computers to train, and AI has also succeeded in setting a new, higher bar for them, even as it disrupted the game. Professor Nam Chi-hyung, who had been teaching Go for more than 20 years, told CNN that AI had became essential in her lessons. “AI can pick the right moves but cannot explain why. People still need me to interpret AI,” she said. And fans use AI viewers to understand who is winning or losing during a match broadcast.

Doesn't all of the above sound very familiar, chess lovers?