谷歌DeepMind的首席执行官Demis Hassabis近期在接受《连线》(WIRED)杂志采访时表达了他的观点，认为人工智能（AI）领域尚未迎来最重大的突破。尽管OpenAI的ChatGPT近期在科技界引起了广泛关注，但Hassabis认为，AI的未来超越了单纯的硬件提升，比如芯片技术的进步。
Hassabis指出，要实现通用人工智能（AGI）的愿景，我们需要的不仅仅是计算能力的极大提升。他强调了在现有技术基础上进行创新的重要性，这包括代理学习、强化学习和规划等领域的持续研究。他提到了DeepMind在AlphaGo项目中就已经开始的这些工作，这个项目成功地让人工智能在围棋这种复杂游戏中击败了世界冠军，展示了AI在学习和策略规划上的潜力。
Hassabis的言论暗示，AI的发展将是一个多元化的进程，需要跨领域的综合进步，而不仅仅是硬件层面的竞赛。他的观点为AI研究提供了新的视角，暗示未来的研究方向可能更加注重算法和学习机制的创新，以推动AI向着更智能、更通用的方向发展。
英语如下：
**News Title:** “DeepMind CEO Envisions AI Future: Breakthroughs Go Beyond Chips, General Intelligence Requires More Innovation”
**Keywords:** AI Breakthroughs, DeepMind, General Artificial Intelligence
**News Content:** Google’s DeepMind CEO, Demis Hassabis, recently shared his perspective with _WIRED_ magazine, suggesting that the field of artificial intelligence (AI) has yet to witness its most significant breakthroughs. While OpenAI’s ChatGPT has recently grabbed headlines in the tech world, Hassabis believes that the future of AI extends beyond mere hardware advancements, such as improvements in chip technology.
Hassabis emphasized that realizing the vision of artificial general intelligence (AGI) requires more than just a tremendous boost in computational power. He stressed the importance of innovation atop existing technologies, including ongoing research in areas like agent learning, reinforcement learning, and planning. He pointed to DeepMind’s work in the AlphaGo project, which began this endeavor and successfully saw AI defeat a world champion in the complex game of Go, demonstrating AI’s potential in learning and strategic planning.
Hassabis’s remarks imply that AI’s evolution will be a multifaceted process, requiring comprehensive progress across disciplines, not just a race in hardware. His perspective offers a new angle for AI research, hinting that future directions may focus more on algorithm and learning mechanism innovations to propel AI towards greater intelligence and universality.
【来源】https://www.wired.com/story/deepmind-ceo-demis-hassabis-interview-artificial-intelligence-scale/
Views: 1