AlphaGo pushed human Go players to be more creative

Earlier this year, an amateur Go player decisively defeated one of the game’s best AI systems. They did this using a strategy devised with the help of a program researcher designed to scan systems like KataGo for vulnerabilities. It turns out the win is just part of a broader Go renaissance that has seen human gamers get more creative since AlphaGO’s milestone win in 2016

In a study recently published in the journal PNAS, researchers from the City University of Hong Kong and Yale found that human Go players have become less predictable in recent years. As the New Scientist explains, the researchers came to this conclusion by analyzing a dataset of more than 5.8 million Go moves made during professional play between 1950 and 2021. Using a “superhuman” Go AI, a program that can play the game and evaluate the quality of each individual move, they created a statistic called the “Decision Quality Index,” or DQI for short.

After assigning a DQI score to each move in their dataset, the team found that the quality of professional play improved relatively little from year to year prior to 2016. The team saw at most a positive mean annual DQI change of 0.2. In some years the overall quality of the game has even dropped. However, since the rise of superhuman AIs in 2018, mean DQI scores have changed at a rate of over 0.7. During the same period, professional gamblers have employed more novel strategies. In 2018, players saw a combination of moves in 88 percent of games, up from 63 percent in 2015, that had not previously been observed.

READ :  FlightSimExpo 2023 already has 50 confirmed exhibitors and sponsors: PMDG, iniBuilds, Navigraph, Orbx and more

“Our results suggest that the development of superhuman AI programs may have prompted human players to deviate from traditional strategies and push them to explore new moves, which in turn may have improved their decision-making,” the team writes .

It’s an interesting change, but not exactly an unintuitive one when you think about it. Professor Stuart Russel of the University of California, Berkeley told the New Scientist, “It’s not surprising that players who train against machines tend to make more moves that machines approve of.”

All products recommended by Engadget are selected by our editorial team independently from our parent company. Some of our stories contain affiliate links. If you buy something through one of these links, we may receive an affiliate commission. All prices are correct at time of publication.