Google’s AlphaGo Beats World Champion In Third Match To Win Entire Series

Go-ing for the sweep

In it’s third consecutive win, Google’s AlphaGo has bested 18-time Go world champion Lee Se-dol. This marks overall defeat for Se-dol in the tournament, despite the fact that the final two games will be played as scheduled.

Google DeepMind will be awarded the $1 million dollar prize, which will then be donated to charity.

AlphaGo’s victory is a watershed moment in artificial intelligence research not because it has practical use in Google’s software offerings, but because it shows how skilled the company has become at tackling difficult problems.

The game of Go was seen to be a nearly insurmountable challenge, because of its sheer complexity. The Go board is a 19×19 grid, and each game has trillions and trillions of potential outcomes.

AlphaGo is a tool built for a very specific job, like an immensely complex wrench. This match demonstrate’s Google’s ability to make an incredible wrench, and the assumption is that knowledge will be carried into other more practical uses somewhere down the line.

While the algorithm behind AlphaGo is different in construction, the display easily mirrors IBM’s 1997 chess series against Garry Kasparov. These public displays that pit man against machine serve as important benchmarks for those who might not follow A.I. research. However, AlphaGo did also serve as a surprise to researchers in the field as well, as most expected an algorithm this complex to be at least 5 years away.

Google DeepMind’s tackles artificial intelligence research from the perspective that if we can train algorithms to perform well in virtual environments, like video games, they would be able to perform similarly in our physical world. In the past few months, the company has released footage of its algorithms playing 3D games, where it successfully navigates mazes to find rewards.

It seems as if this will be the last board game that Google DeepMind tries to tackle in such a public forum. In interviews surrounding AlphaGo’s matches, DeepMind founder and CEO Demis Hassabis has mentioned StarCraft as a game worth pursuing in further research.