Chinese engineers want to challenge Google’s AlphaGo with their own artificial intelligence system that would be better at playing the ancient game than Google’s computer program.
According to the Chinese state-owned Shanghai Security News agency, scientists from the China Computer Go team plan to issue a challenge to AlphaGo by the end of 2016.
AlphaGo has gained fame by beating one of the world’s best players of the strategy board game earlier this month. South Korean professional Go player Lee Se-dol lost to AlphaGo 1 to 4, which marked the first instance when an artificial intelligence system won over a professional Go player. Previously, experts had estimated that artificial intelligence would keep lagging behind the best human players for at least another decade.
The Chinese researchers announced the plan to beat AlphaGo during a conference organised by the Chinese Go Association and the Chinese Association for Artificial Intelligence. However, the participants in the event, called 'The Forum for Understanding the AlphaGo War between Man and Machine and Chinese Artificial Intelligence', did not provide any further details.
Until Google’s subsidiary DeepMind stormed onto the scene with AlphaGo, artificial intelligence systems were at best able to achieve amateur 5-dan level at Go and were unable to beat a professional Go player without handicaps.
Go is considered much more difficult for an artificial intelligence system to master than other board games, such as chess, as it has a much larger number of possibilities at each move. For this reason, traditional artificial intelligence methods struggled with the game.
AlphaGo, in development since 2014, uses an algorithm called Monte Carlo tree search, which decides its moves based on what it has learned previously during training.
The game, most popular in China, South Korea and Japan, involves two contestants moving black and white stones across a square grid, aiming to seize the most territory.