Competitive RTS where DeepMind's AlphaStar reached Grandmaster level using deep learning.

Description

StarCraft II became a major benchmark for artificial intelligence when DeepMind's AlphaStar reached Grandmaster level. Real-Time Strategy requires micro-management, macro-strategy, scouting, and timing. AlphaStar was trained using supervised learning from replays and self-play, eventually defeating professional players. Restrictions were implemented for fairness, such as camera limitations similar to human players. The game features three races with unique units and strategies and a massive competitive multiplayer scene. It hosts a professional esports circuit with global tournaments and includes a single-player campaign. Built-in AI is available for practice with difficulty levels ranging from Very Easy to Insane. Developed by Blizzard Entertainment as the sequel to the legendary original StarCraft, it boasts an especially competitive Korean scene. It is widely used as an AI research platform due to its complexity and high skill ceiling, demonstrating that AI can compete in real-time strategic games. Multiplayer is free-to-play, while campaigns are paid.