By Kim Tae-gyu
When Google DeepMind’s artificial intelligence (AI) program AlphaGo crushed gifted Korean go player Lee Se-dol last March, the world was surprised but some said that a few names better than Lee would be different.
In particular, the world’s two best players, Ke Jie of China and Park Jung-hwan of Korea, were on the lips of observers who were skeptical about the AI’s ability in the Asian board game.
However, Google has silenced doubters once and for all because its AlphaGo program has defeated all human champions, including Ke and Park, in online matches of late.
In fact, users registered with two IDs of “Master” and “Magister” dominated two Korean and Chinese go platforms during the past weeks by scoring convincing victories over all outstanding go players.
Because its style was similar to AlphaGo, people suspected that they would be new versions of AlphaGo, and DeepMind founder Demis Hassabis admitted earlier this week that they were.
“We’ve been hard at work improving AlphaGo, and over the past few days we’ve played some unofficial online games at fast time controls with our new prototype version, to check that it’s working as well as we hoped,” he said.
After the AlphaGo’s landslide 4-1 victory against Lee last March, DeepMind has improved it at breakneck speed as Prof. Jeong Jae-seung at Korea Advanced Institute of Science and Technology (KAIST) demonstrates.
“I took part in a conference at Silicon Valley early December to learn that there are already many algorithms much better than those of AlphaGo,” said Prof. Jeong, a renowned AI expert.
“Just nine months after AlphaGo stunned the world, it has already become a legacy.”
Against this backdrop, pessimism says that humans would never beat AI in go and eventually most other areas.
Ke, who downplayed AlphaGo’s capabilities last year, acknowledged this time that human beings are now no match for AI in the board game that originated in East Asia millennia ago. He lost three times to the AlphaGo in disguise.
Of note is that other countries are also channeling their funds and energy into the potential-loaded AI.
And when it comes to go, Tokyo University came up with a Japanese version of AlphaGo, dubbed Deep Zen Go, last year to take on a Japanese champion Cho Chi-hun. It beat him once in a best-of-three series midway through last November.