Alphabet’s DeepMind AI program AlphaGo takes down ‘Go’ champ
Score another one for the machines! Alphabet’s DeepMind AI research lab has created a program cable of beating a champion in a strategy board game considered much more complex than chess.
The AplhaGo platform earned a clean 5-0 sweep playing against the European Go champ Fan Hui, marking the first time a program has defeated a professional player.
The crushing victory comes in a game where players aim to claim the most territory, by moving black and white round stones around on a square grid.
While machines have been whooping chess pros since IBM’s Deep Blue beat Gary Kasparov 20 years ago, Go was considered more of a challenge for AI programs due to the complex elements of strategy and the greater number of potential scenarios.
In the journal Nature (via Bits), the DeepMind team wrote: “The game of Go has long been viewed as the most challenging of the classic games for artificial intelligence owing to its enormous search space and the difficulty of evaluating board positions and moves.”
The AlphaGo program, which was fed with immense data sets in order to achieve the victory, has also earned a 99.8 per cent win rate against other Go programmes.
However, its victory against a pro player was thought to be decades away. Apparently, the match up with Hui took place in October and since then the machine has continued to improve.
“The machine has continued to get better; we haven’t hit any kind of ceiling yet on performance,” said DeepMind founder Demis Hassabis.
Next up is the world champ with a million bucks on the line. Go humanity!