AlphaGo wins for the third time. (Google via Getty Images)

AlphaGo Vs Lee Sedol: 3-0! Google Deepmind’s AI Impresses Tech World

AlphaGo has made a new record after beating multiple world champion Lee Sedol in a Go exhibition match in South Korea. AlphaGo got three consecutive wins.

The landslide win made the Go world champion feel “powerless.” Sedol said during the conference after the game that he underestimated the capabilities of AlphaGo. He added that he could not deal with the amount of pressure given by the AI.

Lee Sedol also faced the same pressure for the past two games. In game two, he was at the stage where he was understanding the great potential of AlphaGo. Though Lee Sedol made a lead during Game 2’s opening, AlphaGo quickly turned things around.

“There was not a moment in time where I felt that I was leading the game.”

Though Lee Sedol admitted that he missed multiple opportunities throughout the game series, he said that he knew he lost the game “because I wasn’t able to find any weaknesses.”

SEOUL, SOUTH KOREA - MARCH 09: In this handout image provided by Google, South Korean professional Go player Lee Se-Dol (R) puts the first stone against Google's artificial intelligence program, AlphaGo, during the Google DeepMind Challenge Match on March 9, 2016 in Seoul, South Korea. Lee Se-dol played a five-game match against a computer program developed by a Google, AlphaGo. (Photo by Google via Getty Images)
[Photo by Getty Images Handout]

In an analysis of Game 3, 9-dan Go master Yoo Changhyuk noted that Lee Sedol has successfully implemented challenging strategies to fight AlphaGo.

“During the first match, Lee Sedol made difficult moves to agitate AlphaGo, but failed to do so. Today, he tried the opposite — he played safe and entered the endgame.”

Lee Sedol will have two more chances to fight AlphaGo since they have to complete 5 games to finalize the scores.

AI Leveling the Playing Field

Deepmind CEO Demis Hassabis was “stunned and speechless” with the results. In a paper published on Nature, it was explained that AlphaGo was programmed to understand and mimic the strategic analysis of expert Go players in the world. The paper also noted that AlphaGo has a capability to expand its knowledge by just playing a game against itself.

“Here we introduce a new approach to computer Go that uses ‘value networks’ to evaluate board positions and ‘policy networks’ to select moves. These deep neural networks are trained by a novel combination of supervised learning from human expert games, and reinforcement learning from games of self-play. Without any lookahead search, the neural networks play Go at the level of state-of-the-art Monte Carlo tree search programs that simulate thousands of random games of self-play.”

After AlphaGo won the match last Mar. 10, SpaceX founder Elon Musk congratulated DeepMind for the feat.

“Many experts in the field thought AI was 10 years away from achieving this,” said the international tech leader.

DeepMind programmer David Silver explained that AlphaGo uses a different “value network.”

“Instead of having this enormously deep search that has to go all the way down to perhaps 300 moves, all the way down to the end of the game, what we do is we search to some modest depth, of perhaps 20 moves, and then we evaluate that position, without playing all the way to the end of the game,” Silver explains.

Demis Hassabis, CEO of Google's artificial intelligence (AI) startup DeepMind, speaks during a press conference on March 8, 2016 in Seoul, South Korea. Lee Se-dol is set to play a five-game match against a computer program developed by a Google, AlphaGo, starting March 9. (Photo by Kim Hee-Chul-Pool/Getty Images)
[Photo by Kim Hee-Chul/Getty Images Pool]

Hassabis told the press that the match series is very important to their research. Though AlphaGo has already showed the world that AI possibilities are real and achievable, they needed to test it out in the real world to see its true form and also explore new ways of improving it.

Hassabis also noted that Lee Sedol has been a valuable part of AlphaGo’s exploration. Though AlphaGo can do internal plays, they “need somebody of the incredible skill of Lee Sedol to creatively explore and see what weaknesses AlphaGo maybe has, so we can see them. That’s why we’re having this match to find out,” said Hassabis.

In an interview with The Verge, Hassabis said that AI vs. man showdown could be a possibility in the future. Ultimately, for the Deepmind team and Hassabis, Alpha go have become the “holy grail for AI research.”

AlphaGo and Lee Sedol have two matches left. Watch the full coverage of Match 3 below.

[Photo by Google via Getty Images]