AlphaStar AI beats 99.8 percent of all Starcraft 2 players

In July we had already reported that Google’s learning-based Starcraft II AI Alphastar was to compete against real players in an experiment for quite some time. The experiment was then carried out and the players did not know that they were playing against an artificial intelligence and not against a real human. Now Google reports in a blog entry also the result of the experiment. According to this, Alphastar was able to achieve a victory in 99.8 percent of all games.

Google had already demonstrated the power of the Alphastar AI developed by Google’s Deepmind company earlier this year when the AI managed to beat two well-known Starcraft 2 professionals.

At that time, the AI player “Alphastar” controlled by Alphastar was able to beat the pros named TLO and MaNa in ten consecutive Starcraft 2 games. Only in an eleventh game were the pro players able to somewhat mitigate the ignominy and win. The bottom line was that the AI won 10 to 1, although there were accusations that the rules in place during the games favored the AI player.

In the new announcement, Google points out that Alphastar has now successfully held its own without any game restrictions in the real-time strategy game Starcraft II, which is extremely popular in the e-sports sector. Full games had been played and under “professionally recognized conditions”. All games took place in the months of July and August 2019.

Accordingly, all of the rules that apply to human players would have applied to Alphastar, such as restrictions on how you can see what is happening in a level or limits on how many actions you can take in a second. In the 1-on-1 player games, Alphastar also competed in all three Starcraft 2 races, Protoss, Terran or Zerg. In addition, all games took place on official game servers via Blizzard’s Battle.net.

The result: As mentioned above, Alphastar won 99.8% of all games. Alphastar also reached the Grandmaster level in all three Starcraft 2 races.

In the blog post, Google researchers also point out that the learning-based AI insights gathered with Starcraft 2 can be applied in many other areas as well. The biggest challenge for Alphastar, they say, was figuring out for itself the right strategies that would lead to victory by constantly playing Starcraft 2. “AlphaStar is a fascinating and unorthodox player – one with the reflexes and speed of the best professionals, but strategies and a style that is completely unique,” it says. It continues, “The way AlphaStar has been trained (…) has resulted in gameplay that is unimaginably unusual; it really makes you wonder how much of the wide range of possibilities StarCraft offers pro players has really been explored.”

Follow us and check out our social media accounts on Twitter, Facebook & YouTube ►

● on Twitter ► esport.directory
● Facebook ► esport.directory
● Youtube ► esport.directory

StarCrafttips, StarCrafttricks, StarCraftcheats, StarCrafthacks, StarCrafthacks2022, StarCrafttipspro, StarCraftpaidcheats, StarCraft, StarCrafttipsandtricks, StarCrafttipsandtricks2022, StarCrafttipsandtricksforbeginners, StarCrafttips, StarCrafttricks, StarCrafttrickshots, StarCrafthacks, StarCrafthacksfree, StarCrafthacksfree2022,