Developed by a Chinese AI team, Golaxy has a very different algorithm than AlphaGo, Google's Go program that defeated Ke last year, said Jin Xing, chairman of Golaxy.

"We are exploring new methods to consume fewer computing resources and training samples," Jin said. "Golaxy has made progresses in the feature system, model structure and MCTS algorithm, compared with AlphaGo."

The results of human vs. AI Go games are no longer suspenseful, said Lin Jianchao, chairman of the Chinese Go Association.

The upcoming competition between Ke Jie and Golaxy will further test the competitive ability of the domestic AI Go program, and it will be a starting point for wide use of AI go technology in public apps, Lin said.

AlphaGoZero, the upgraded version of AlphaGo, benefits from a large number of self-play, and iteratively saves and upgrades the models.