Google DeepMind to Train AI Models on Eve Online and Invest in Fenris Creations
Google DeepMind disclosed Wednesday that it will tap the massive multiplayer online role‑playing game Eve Online as a sandbox for training its AI systems. The move marks the first time the AI lab has partnered with an active, player‑driven universe that spans more than two decades of space combat, piracy and commerce.
Eve Online, created by Icelandic developer CCP Games, launched 23 years ago and has built a devoted community that conducts large‑scale battles, mines resources and runs virtual corporations. Those activities generate in‑game assets worth hundreds of thousands of dollars, creating high‑stakes decision‑making that DeepMind hopes to analyze.
DeepMind’s chief executive, Demis Hassabis, explained that games have long served as testbeds for the lab’s breakthroughs, from Atari DQN to AlphaStar. "They’re the perfect training ground for developing and testing AI algorithms," he said in a statement. The company plans to run experiments on isolated servers so live gameplay remains unaffected.
Alongside the research partnership, DeepMind is taking a minority stake in Fenris Creations, the new name for CCP Games after its recent ownership change. The investment, described by Fenris Creations CEO Hilmar Veigar Pétursson as being "in the millions of dollars," will give the AI lab a seat at the table without altering the studio’s independence. Fenris Creations will retain its own board of directors, allowing it to make the "far‑reaching decisions" required to support Eve’s complex ecosystem.
The stake comes after the game’s previous owner, South Korea’s Pearl Abyss, sold CCP back to its management last week. DeepMind’s involvement, according to a company blog post, is intended to advance AI research safely within a player‑driven universe as intricate as Eve Online. Over time, the research could also feed back into the game, helping developers refine mechanics and enhance player experiences.
Industry observers note that Eve Online’s emergent narratives and prolonged battles provide a rich dataset for modeling human behavior, cooperation and competition. By analyzing how thousands of players interact on a massive scale, DeepMind hopes to improve its ability to predict and respond to complex, real‑world scenarios.
The partnership underscores a growing trend of AI labs turning to interactive digital environments for training data. As DeepMind leverages Eve’s virtual economy and combat systems, the collaboration could set a precedent for future AI‑gaming alliances.
Used: News Factory APP - news discovery and automation - ChatGPT for Business