Google DeepMind Unveils More Powerful AlphaGO Model, Beats World Champion
There's a mystery player in the world of competitive Go. The complex strategy game is extremely hard to master, as it almost has an infinite number of moves. Chess masters be damned, it appears the mysterious player is actually yet another robot.
According to Nature, Google DeepMind has recently revealed its newest mystery player is an updated version of Alpha Go.
It's only known as Master(P), and this anonymous player has beaten the world's best in the game in a series of online games. It even defeated current number one Ke Jie, who is just 19 years old.
According to Scientific American, the game's complexity has been famously difficult to crack even for computers. Apparently, this is not the case, as early last year, the Go world was stunned with the arrival of AlphaGo. It defeated a professional human player, Fan Hui, and then beat one of the game's top players, Lee Sedoi.
Fellow players already had a hunch that Master(P) may be an AI. It came out of nowhere to just win dozens of consecutive quick-fire skirmishes on two separate online platforms.
On the fourth of January this year, Demis Hassabis, Google DeepMind chief executive, revealed that Master(P) is actually the newest prototype of AlphaGo.The unofficial games were designed to test the prototype.
According to Nature, it played on the servers Tygem and FoxGo. It played more than 50 games and won it all, except for the one game because the network connection of Chen Yaoye timed out. After losing to Master(P), Chinese professional Gu Li offered US $14,400 to any human who could defeat the mysterious player. But to no avail.
A lot of observers had the initial suspicion that maybe another team had created an AI that could master the game, as both China and South Korea were already trying this for a time.
The newest version of AlphaGo would play official, full-length games this year. Of course how strong it will be in more high-profile tournaments is still unclear.