Nerd is the Word: One step closer to human-like robots
IF you've played Go then you would know it requires a lot of skill and a lot of intuition to play.
Computers aren't well-known for their ability to predict human intuition and that's why artificial intelligence has never been able to beat Go professional players.
But two decades after computers first beat professional chess players, Google subsidiary DeepMind has created a program that can beat some of the best Go players.
The concept of the 2500 year old game Go is simple, but it's difficult to master. The range of possible moves makes prediction very difficult.
The game is often compared to chess but rather than playing on an 8x8 board, you play on a 19x19 board.
Players must place black and white stones on the board with at least one empty space next to it, or part of a group of stones of the same colour.
If a player's stones are surrounded by an opposing stone or group of stones it is "captured".
The objective is to have surrounded the largest total area of the board.
DeepMind's Go-mastering program, AlphaGo, uses very clever algorithms to master the game, rather than to simply win the game.
It was trained by observing millions of boards of Go and using these observations to build up a predictive model.
This was then back up by reinforcement learning. AlphaGo played against itself, learning from its victories and losses.
DeepMind did not create this program simply to beat the world's best Go players, either.
This breakthrough points to a world where robots can learn to perform tasks and respond to their environment.
The creators hope their algorithms can be one day used to address some of society's biggest issues, such as climate modeling.
For now, the creators of AlphaGo are focused on their final battle - to win a match against the Go world champion, Lee Se-dol.
Lee said in the scientific journal Nature that "regardless of the result, it will be a meaningful event in the baduk (Go) history".
"I heard Google DeepMind's AI is surprisingly strong and getting stronger but I am confident that I can win at least this time," he said.
The match will be live-broadcasted in March so we will have to wait and see.