Computer Go Champ Flattens Human Opponent

On This Site

Current Events

Share This Page






Follow This Site

Follow SocStudies4Kids on Twitter

May 28, 2017

Score another win for AI. AlphaGo, a product of Google's DeepMind project, has defeated the world's top Go player in three straight matches. The victory followed a similar triumph a year earlier, against another top human player.

This time, it was China's Ke Jie showing his Go ability. He has been the world's top Go player for more than two years, but he still lost out in three straight contests against the computerized opponent.

The matches took place in Wuzhen, China. AlphaGo won the first match by just one half-point, the closest possible margin of victory. As Ke would attest to, though, a victory is a victory. As well, Google's programmers were quick to stress, the computer engine favors moves that are more likely to guarantee victory and so AlphaGo won't run up the score on an opponent, computerized or otherwise.

Seeing the writing on the wall, Ke resigned from the second game, handing the AI the technical 2-out-of-three victory. The third game was no better for Ke.

Both players had studied the other's moves, and Ke even tried some unconventional moves that he said later he would not have used against a human opponent.

Ke had already suffered defeat twice at the hands of AlphaGo, whose identity was concealed in an online competition in January. AlphaGo, going by the name of Master, won 50 online games in a row. 

In 2016, the human opponent was South Korea's Lee Sedol, the then-world champion who lost the first three and the fifth of five matches but defeated AlphaGo in the fourth match.

Go has been played for more than 2,500 years. Two players (one playing white stones and the other playing black stones) commonly play on a board that is a square grid made up of 19 squares on a side. (Some boards are smaller.) Players place stones on the intersection between squares. The object is to claim the most territory on the board by surrounding the stones of the other player. 

In a game of Go, a player commonly has a potential of about 200 moves, making the game extremely complex. Chess (also mastered by a computer), by contrast, has a few dozen moves possible at most opportunities. 

People involved in the DeepMind project are already working with the United Kingdom's National Health Service, in a post-gaming move similar to that pursued by IBM's Watson, after it mastered the TV show Jeopardy! in 2011.

DeepMind has also said that it would release all of the data mined from the three-game series and would also work with Ke on a teaching tool.

Search This Site

Custom Search

Social Studies for Kids
copyright 2002–2017
David White