谷歌DeepMind近日宣布推出了一项名为SIMA（可扩展、可指导、多世界智能体）的新型人工智能技术，该技术旨在让AI学习并掌握游戏技能，使其游戏表现更接近人类玩家。SIMA智能体代表了人工智能在模仿人类行为和适应复杂环境方面的新突破，目前仍处于研发阶段。
据《The Verge》报道，SIMA已经在《Valheim：英灵神殿》、《无人深空》和《模拟山羊》等多款不同类型的游戏上进行了训练。通过这些训练，SIMA不仅能够学习游戏规则，还能适应各种游戏环境，包括那些没有明确终点或线性路径的开放世界游戏。这表明SIMA的智能具有高度的可扩展性和适应性，能够处理游戏世界的多样性和不确定性。
谷歌DeepMind的这一创新旨在推动AI在游戏领域的应用，未来可能有助于游戏设计、测试和玩家体验的提升。通过让AI学习如何像人一样玩游戏，科研人员可以更好地理解人类的学习过程，并可能将这些洞察应用到更广泛的领域，如机器人技术、自动驾驶和复杂决策系统。
然而，尽管SIMA展现了巨大的潜力，但其全面应用仍需克服诸多挑战，包括如何在更广泛的游戏中保持其学习效率和灵活性，以及如何确保AI在模仿人类行为时保持道德和伦理的正确性。谷歌DeepMind将继续进行深入研究，以期将SIMA的智能体技术推向新的高度。
英语如下：
News Title: “Google DeepMind Unveils SIMA: A New AI Breakthrough in Learning to Play Games Like Humans”
Keywords: Google DeepMind, SIMA Agent, Game Training
News Content: Google DeepMind recently announced the development of a novel artificial intelligence technology called SIMA (Scalable, Instructable, Multi-World Agent), designed to enable AI to learn and master gaming skills, emulating human players. SIMA represents a significant breakthrough in AI’s ability to mimic human behavior and adapt to complex environments, though it remains in the research phase.
As reported by _The Verge_, SIMA has been trained across a variety of games, including _Valheim:英灵神殿_, _No Man’s Sky_, and _ Goat Simulator_. Through this training, SIMA not only grasps game mechanics but also adapts to diverse in-game environments, including open-world games with no clear endpoints or linear paths. This demonstrates the high scalability and adaptability of SIMA’s intelligence, capable of handling the variability and uncertainty of gaming worlds.
Google DeepMind’s innovation aims to propel AI applications in the gaming industry, potentially benefiting game design, testing, and player experience in the future. By teaching AI to play games as humans do, researchers can gain deeper insights into the human learning process, which could be applied to broader fields such as robotics, autonomous driving, and complex decision-making systems.
However, despite its immense potential, widespread application of SIMA still faces challenges, including maintaining its learning efficiency and flexibility across a broader range of games and ensuring ethical correctness as AI mimics human behavior. Google DeepMind will continue its in-depth research to push SIMA’s agent technology to new heights.
【来源】https://www.theverge.com/2024/3/13/24099024/google-deepmind-ai-agent-sima-video-games
Views: 2