Face AlphaStar AI on StarCraft II ladder
Face AlphaStar on StarCraft II ladder ⚡⚡⚡ Esports and gaming news, analytics, reviews on WePlay! The latest news on WePlay!
DeepMind will be giving all StarCraft II fans the chance to defeat their program named AlphaStar. With the desire to improve AI research, AlphaStar was created to learn how to balance short and long-term goals and adapt to unexpected situations posed by StarCraft II and its players.
"We’re excited to offer our community a chance to contribute to this process by playing against AlphaStar in a small number of blind trial games that will allow DeepMind to benchmark its agents."
-Blizzard, StarCraft II blog
After becoming the first AI to defeat a professional player in December 2018, DeepMind will take their research even further by pitting experimental versions of AlphaStar against players on StarCraft II's ladder. Taking part is as easy as clicking the "opt-in" button located on an in-game pop-up window. Players can also change their opt-in status through the “DeepMind opt-in” button situated in the 1v1 Versus menu.
Players that opt-in and meet AlphaStar on the ladder will have no indication of the fact, allowing DeepMind to benchmark its performance under the same conditions.
The AI will be treated like any other player, with any win or loss against AlphaStar affecting MMR as it normally would. Matchmaking will also occur like normal, with AlphaStar playing as and against Terran, Zerg or Protoss.
In a bid to keep things fair, AlphaStar will have its actions per minute and per second capped so it doesn't exceed human capabilities. This restriction will give players a better chance at winning than what was available during past demonstrations.
"AlphaStar perceives the game using a camera-like view. This means that AlphaStar doesn’t receive information about its opponent unless it is within the camera’s field of view, and it can only move units to locations within its view. All limits on AlphaStar’s performance were designed in consultation with pro players."
DeepMind hopes to gather a wide range of results from this test. At its conclusion, they will then release its results in a peer-reviewed scientific paper, together with replays of AlphaStar's matches.