Deep Mind Artificial Intelligence Wins 10 Out 11 Games Versus Pro Human Players

Deep Mind has a StarCraft II program AlphaStar. It is the first Artificial Intelligence to defeat a top professional human player.

In a series of test matches held on 19 December, AlphaStar decisively beat Team Liquid’s Grzegorz “MaNa” Komincz, one of the world’s strongest professional StarCraft players, 5-0, following a successful benchmark match against his team-mate Dario “TLO” Wünsch. The matches took place under professional match conditions on a competitive ladder map and without any game restrictions.

AlphaStar had 10 wins against StarCraft II pros TLO and MaNa in two separate five-game series in December. MaNa took on AlphaStar in a live match streamed by Blizzard and DeepMind and was able to win.

In MaNa win AlphaStar was not given an overhead view. It had a human player view. MaNa was able to exploit some of the AI’s shortcomings and win. This was AlphaStar first loss against pro players.

AlphaStar had unique strategies each time it played. The AlphaStar games are available for replay viewing so humans can learn AlphaStar strategies.

AlphaStar and Deep Mind generated simulated players from previous human pro games. They created a series of virtual games which was called the AlphaStar League. The AlphaStar League ran for one week and the matches produced new information that helped to refine the AI’s strategy. Over one week, AlphaStar played the equivalent of 200 years worth of StarCraft II. DeepMind chose five individual agents that they determined had the least exploitable strategy and had the best chance to win. Five agents were used to produce the five-game sweeps.

Although there have been significant successes in video games such as Atari, Mario, Quake III Arena Capture the Flag, and Dota 2, until now, AI techniques have struggled to cope with the complexity of StarCraft.

AlphaStar plays the full game of StarCraft II, using a deep neural network that is trained directly from raw game data by supervised learning and reinforcement learning.

Starcraft is an AI Grand Challenge

StarCraft is considered to be one of the most challenging Real-Time Strategy (RTS) games. It is considered as a “grand challenge” for AI research.

StarCraft II was created by Blizzard Entertainment. It is set in a fictional sci-fi universe and features rich, multi-layered gameplay designed to challenge human intellect. It is among the biggest and most successful games of all time. Players have competed in esports tournaments for more than 20 years.

There are several different ways to play the game, but in esports the most common is a 1v1 tournament played over five games. To start, a player must choose to play one of three different alien “races” – Zerg, Protoss or Terran, all of which have distinctive characteristics and abilities.