The Significance of AlphaGo’s Victory

Back to Article
Back to Article

The Significance of AlphaGo’s Victory

Jay Park, Staff Writer

Hang on for a minute...we're trying to find some more stories you might like.

Email This Story

Lee Sedol, the strongest Go player in the world, lost 1-4 against AlphaGo, an artificial intelligence (AI) developed by Google DeepMind’s project team. It was a loss that was unimaginable prior to this significant match. It broadcasted on all major South Korean networks and accumulated an audience of 60 million in China.

  Just a few years ago, the prospect of a machine beating a top human player in Go was unthinkable. The computer has conquered humans in chess, backgammon, poker, and even Jeopardy, but not Go.

  Go, or its other names Baduk and Weiqi, is the eastern version of chess. It is laid out on a 19-by-19 board, where two players take turns placing white and black stones respectively on the intersection of the lines. The game is based around surrounding your opponent’s stones and claiming their “territory” of stones. In order to win, one must claim more total stones than his or her opponent.

  This means that the possibilities for Go span far more than chess with 1.01^7 outcomes, which is miniscule compared to the 2.08^59 different outcomes of Go. The overwhelming numbers made it nearly impossible for the computer to be able to beat human intelligence. Despite these difficult odds, AlphaGo appeared to easily trounce the century’s best Go player.

  An hour before the matches, Lee Sedol claimed that he would win 5-0 or 4-1 against AlphaGo. His confidence stemmed from observing how the AI played against the match against its previous opponent Han Fui, the European Go Champion. Han Fui soundly lost 5-0 against AlphaGo. Fui’s loss was certainly unexpected, but Lee Sedol knew he was better than Fui. Sedol’s complacency led to a false sense of security, thus he failed to realize that in the five months of the match against Han Fui, AlphaGo was self-improving exponentially.

  AlphaGo won the first three matches, deciding the outcome of the set. However, as a tradition to play all five matches, the competition continued. By exploiting the AI’s single mistake at turn 87, Lee Sedol won the fourth match. This was more than just a mistake in the long run. The victory proved that the AI was not infallible and had long ways to go before perfection. However, AlphaGo returned to its flawless play in the fifth game and secured yet another victory. Lee Sedol later talked about his embarrassment over some of his mistakes during interviews, but stated that he respected the AI’s meticulous moves and adaptability.

  The project AlphaGo was the collaboration of a 100 scientists. The reason behind the project was to test the capability of deepmindDeepMind’s artificial intelligence. It was decided that complex games were apt for developing flexible and adaptable algorithms. Success in this area means that machines could possess the ability to tackle problems by thinking like humans. The AI used reinforcement learning to play millions of games against itself, constantly improving its own database. The longer it existed, the more it could learn.

  What does this mean for us? For starters, this means that the science fiction scenario of robots taking over jobs is not too far-fetched. Artificial intelligence could replace certain tasks, such as accounting. Even lawyers, who undergo extensive education, would be impacted. Machines could process legal documents and return a proper calculated response instantly. Much of what a lawyer could do would be undermined. In a similar aspect, any task that involves unidirectional work could be automated.

  In fact, many of the existing jobs might disappear, but new jobs will be created. Speculations predict jobs such as memory augmentation surgeon and nano-medic will exist in order to harness new technology. Eventually, all forms of work will necessitate the use of technology because the speed and accuracy of a computer holds a major advantage beyond an average human capabilities.

  However, artificial intelligence is still in its infantile stages, proven by how Lee Sedol scored a victory over its mistake in game four. While not perfect, we as the upcoming generation and workforce, should keep an eye on its rapid progress.

Print Friendly, PDF & Email