{"id":149153,"date":"2026-08-24T08:10:12","date_gmt":"2026-08-24T08:10:12","guid":{"rendered":"https:\/\/www.europesays.com\/ai\/149153\/"},"modified":"2026-08-24T08:10:12","modified_gmt":"2026-08-24T08:10:12","slug":"google-deepmind-to-conduct-ai-research-on-eve-online-utilizing-the-virtual-universe-which-has-been-around-for-over-20-years-as-a-mirror-reflecting-intelligence","status":"publish","type":"post","link":"https:\/\/www.europesays.com\/ai\/149153\/","title":{"rendered":"Google DeepMind to conduct AI research on ‘EVE Online,’ utilizing the virtual universe, which has been around for over 20 years, as a ‘mirror reflecting intelligence.’"},"content":{"rendered":"

Aug 24, 2026 16:15:00<\/p>\n

<\/a><\/p>\n

Google DeepMind has revealed specific details about its research partnership with Fenris Creations, the developer of the space MMORPG ‘EVE Online,’ including how the EVE series will be used for AI research. They will use the world of EVE Online, where players have built economies, wars, and diplomacies for over 20 years, to research technologies that enable AI to learn, remember, and plan while interacting with humans over long periods of time.<\/p>\n

Exploring new frontiers of AI and games research \u2014 Google DeepMind<\/p>\n

https:\/\/deepmind.google\/blog\/from-atari-to-eve-online-building-on-15-years-of-ai-research-in-games\/<\/a><\/p>\n

Since its founding in 2010, DeepMind has used games in its AI research, developing Deep Q-Network (DQN), which learns 49 different Atari 2600 games such as ‘Pong,’ ‘Breakout,’ and ‘Space Invaders’ by only looking at pixels on the screen. Since then, it has developed numerous AI capable of playing games, including AlphaGo, which defeated world-class Go players; AlphaZero, which learned chess, shogi, and Go using the same mechanism; and AlphaStar, which reached Grandmaster level in the real-time strategy game ‘StarCraft II.’<\/p>\n

DeepMind, the developers of the Go AIs ‘AlphaGo’ and ‘DQN,’ is attempting to build the strongest AI using ‘StarCraft 2’ – GIGAZINE
<\/a><\/p>\n

<\/p>\n

Traditional game AI research makes it easy to set clear goals such as ‘winning the game’ or ‘achieving a high score,’ but the real world lacks easily understandable scorecards or reset buttons. Therefore, Google DeepMind is researching AI that can not only complete a given game but also understand its surroundings and act accordingly, just like a human.<\/p>\n

One of the research results is the ‘Scalable Instructable Multiworld Agent (SIMA).’ Instead of accessing game-specific APIs or source code, SIMA looks at the screen like a human player, understands natural language instructions, and operates the keyboard and mouse. ‘SIMA 2,’ which uses Gemini, also supports real-time inference and conversation, and research is progressing in multiple 3D game environments such as ‘No Man’s Sky,’ ‘Valheim,’ and ‘Hydroneer.’<\/p>\n

Google DeepMind announces ‘SIMA 2,’ an AI agent that learns by playing 3D games like a human – GIGAZINE
<\/a><\/p>\n

<\/p>\n

However, many games used in AI research have the characteristic that the situation is reset after each match and the objectives, such as winning or losing, are relatively clear. On the other hand, in EVE Online, which began service in 2003, a single universe shared by many players is maintained over a long period of time, and the economy and resource shortages due to transactions between players, cooperation and conflict between corporations and alliances, and diplomacy are constantly changing. The fact that the game world itself is not easily reset is a major difference for AI research.<\/p>\n

In EVE Online gameplay, you not only need to win battles between spaceships, but also to deduce the intentions of other players from limited information and make decisions while considering betrayal and market fluctuations. Google DeepMind has identified the following abilities as particularly important for EVE Online research: ‘Continuous Learning,’ which is the ability to keep learning without losing previous knowledge even when the environment changes; ‘Memory,’ which is the ability to accumulate information over long periods and retrieve it when needed; ‘Long-Term Planning,’ which is the ability to think several weeks to several years ahead; and the ability to cooperate, compete, and negotiate with multiple opponents.<\/p>\n

According to developer Fenris Creations, they will initially use an offline version of EVE Online running on a local server to conduct research in a controlled environment utilizing historical data and simulations. Fenris Creations explains that their goal is not to create an AI that can play the game on behalf of the player or eliminate risks and consequences.
<\/a><\/p>\n

The research partnership is considering using not only EVE Online but also other titles in the EVE series as a research environment. ‘EVE Vanguard’ requires quick tactical decisions from a first-person perspective, making it suitable for studying rapid decision-making. On the other hand, ‘EVE Frontier’ has ‘Smart Assemblies’ that players can program, which can be used as a place to study adaptation to environments where the world’s mechanics themselves change. Google DeepMind says that by using multiple EVE titles, they can handle decisions across different timeframes, from instantaneous actions to galaxy-scale long-term strategies.<\/p>\n

The collaboration between the two companies extends beyond the research environment and is also reflected in some player-facing features. For example, as a beginner support feature, a system called ‘Aura Guidance’ is being tested as a prototype, which allows new players to resolve their questions within the game based on questions and answers exchanged in actual Rookie Help sessions. If the answer is not entirely certain, it directs the player to Rookie Help sessions with human participants, so it is designed to assist rather than replace the human community.<\/p>\n

Google DeepMind states that ‘games have always been a mirror reflecting intelligence,’ explaining that the more complex and long-lasting the game world becomes, the more advanced the AI acting within it needs to be. In the long term, they aim not only to use research findings from games to create new gaming experiences and improve accessibility, but also to apply the technologies and ideas cultivated in game research to real-world problems and scientific research, citing examples such as AlphaFold.<\/p>\n","protected":false},"excerpt":{"rendered":"Aug 24, 2026 16:15:00 Google DeepMind has revealed specific details about its research partnership with Fenris Creations, the…\n","protected":false},"author":2,"featured_media":149154,"comment_status":"","ping_status":"","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[9],"tags":[24,8440,6526,4376,8439,8441,132,1429,717,130,8437,1110,66,4176,136,8438],"class_list":["post-149153","post","type-post","status-publish","format-standard","has-post-thumbnail","category-google","tag-ai","tag-anime","tag-blog","tag-food","tag-game","tag-gigazine","tag-google","tag-google-ai","tag-hardware","tag-internet","tag-it","tag-mobile","tag-news","tag-note","tag-software","tag-web-service"],"_links":{"self":[{"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/posts\/149153","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/users\/2"}],"replies":[{"embeddable":true,"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/comments?post=149153"}],"version-history":[{"count":0,"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/posts\/149153\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/media\/149154"}],"wp:attachment":[{"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/media?parent=149153"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/categories?post=149153"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/tags?post=149153"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}