Google Achieves Breakthrough In Artificial Intelligence As Program Defeats Europe’s Champion In Chinese Game Go

by Dalal Nasif / Feb 04, 2016 02:35 PM EST
The Chinese game Go is said to be too complex that it has more possible positions than the number of atoms in the universe. (Photo by China Photos/Getty Images)

Google has just advanced itself a step further from other artificial intelligence companies (AI) like Facebook. The tech firm recently achieved a breakthrough in research by defeating the current best European player of the Chinese game Go, which is said to be far more complex than chess.

Google invited Fan Hui, the deemed elite player who devoted his life to Go as early as the age of 12, in its London office last October and pitted the company's own version AlphaGo against him. They played five games. The computer won all of them, according to Korea Joong Ang Daily News.

This level reached by Google has surprised experts who thought that an AI defeating a human professional can only take place in some few decades away.

"Many of the best programmers in the world were asked last year how long it would take for a program to beat a top professional, and most of them were predicting 10-plus years," said Demis Hassabis, Chief Executive of the company's division DeepMind which handled the project.

AI researchers has caught the attention of Go as the game makes it possible for the player to choose from around 30 million moves. While each move in chess has 20 choices per move, Go takes on some 200 options, making it hard to crack for computers.

Google deviated from the other earlier versions of AI specializing in Go sincce the company did not opt for a pre-programmed technology. They instead designed an AI that utilizes learning curves by following a three-stage process of pattern tracking, problem solving and the Monte Carlo Tree Search, as reported in BBC News.

Facebook is also using similar search technology as Google, but the latter's version called AlphaGo achieved higher scores than Facebook's based on the data revealed by both companies.

In Google's official blog post, Hassabis revealed that AlphaGo is set to embark on a journey to Seoul in March where it will be facing the world's grandmaster of Go, Lee Sedol. The winner will receive $1 million.

Moreover, the company's representative said that because the expert system they have built uses a general-purpose method, they are hoping to apply it in the real-world context and "address some of society's toughest and most pressing problems, from climate modelling to complex disease analysis."

© 2024 Korea Portal, All rights reserved. Do not reproduce without permission.

Don't Miss