Site icon TechloMedia

Google DeepMind Partners With EVE Online Developer to Train AI Using the Long-Running MMO

Google DeepMind Partners With EVE Online Develope

Google is taking a minority stake in the company behind EVE Online as part of a new partnership focused on artificial intelligence research. According to reports, Google DeepMind will use EVE Online to train and study advanced AI systems, particularly around long-term planning, decision-making, and adaptive behavior in large-scale virtual environments.

The studio formerly known as CCP Games has also been rebranded as Fenris Creations following its move to reacquire the rights to the EVE franchise from Pearl Abyss. Fenris Creations CEO Hilmar Veigar Pétursson said Google’s investment is worth “millions” of dollars.

It may look like another gaming partnership involving AI. But EVE Online is not just another multiplayer game. The game has been running for more than 20 years and is widely considered one of the most complex online virtual worlds ever created.

Unlike traditional games with scripted paths and predictable objectives, EVE Online operates more like a living digital society. Players manage economies, form political alliances, wage wars, spy on rivals, manipulate markets, and coordinate strategies that can unfold over months or even years. That makes the game extremely valuable for AI research.

Most current AI systems are very good at short-term tasks. They can answer questions, generate images, or react quickly in controlled environments. But long-term reasoning, adaptive planning, and continuous learning remain much harder problems.

According to DeepMind’s Adrian Bolton, EVE Online presents challenges AI still struggles with, including “long-term planning and continual learning.” That is exactly why Google is interested.

This game would help Google’s AI system understand environments that behave more like the real world. EVE Online contains unpredictable human behavior, dynamic economies, cooperation, betrayal, risk management, and evolving social structures. Those are things AI models cannot fully learn from static datasets or traditional benchmarks.

If DeepMind can train AI systems that successfully navigate the complexity of EVE’s universe, those systems could later become better at handling real-world tasks involving planning, negotiation, resource management, and adaptation.

The partnership may also help Google improve AI memory and long-term context retention, which remain major weaknesses in current generative AI systems.

Google previously used games like StarCraft II for AI experiments, but EVE Online is significantly more open-ended and socially driven.

The companies say the partnership will also explore “new gameplay experiences” powered by AI technologies, although no specific features have been announced yet. This could eventually lead to smarter NPCs, adaptive missions, dynamic economies, or AI-driven events inside future EVE titles.

Exit mobile version