On April 27, Ke Jie, the world’s second-ranked Go player competed against another AI Go program after having defeated by AlphaGo at 0-3 last year. (Go is an extremely intensive chess game, especially at higher level gameplays.)
This time, Kejie was playing against the Chinese AI Go Program “Golaxy”. He admitted defeat at the 145th round/step.
In the interview afterwards, “incapability” and “ hopelessness” were the most referred words in his remarks.
“There is always a sense of incapability when I play Go against AI, as its algorithm and overall judgment are far beyond human capabilities. I can’t say my performance is good or not. It’s plainly difficult (for humans) to go up against AI,” said Ke jie in the interview.
He also pointed out that the key is in the middle stage of the game. He did not fall far behind when he was just laying out his general game plan and board structure at the beginning. However, when it comes to the middle stage, AI can forsee moves and strategize much better than humans as the complexity of the game rises.
The technology of “Golaxy” was based on “Abacus,” a GO AI previously developed by Tsinghua University. Golaxy also incorporated some of the structure of AlphaGo, and optimized its performance further with new algorithmic innovations. Prior to the confrontation with Ke Jie, Golaxy has competed against many professional Go players since April 16 this year. Golaxy won 28 out of 30 online competitions with a win rate of 93.3%.
It has become more or less an established notion that AI is superior to human intelligence in the Go scene. However, it doesn’t mean Go – an intellectual sport with over 4000 years of history – has become less meaningful or valuable of a practice due to the introduction of AI. When asked whether he would take part in the “Human V.S. AI” competitions in the future, Ke answered “yes” without hesitation.