Meta 的首席科学家、图灵奖获得者 Yann LeCun 近日在一个深度访谈中向科技博主 Lex Fridman 表达了他对人工智能（AI）安全性的坚定信念。在接受长达近三个小时的访谈中，LeCun 明确表示，他认为 AI 毁灭人类的可能性为零，反驳了普遍存在的 AI 末日论。
LeCun 在讨论中详细剖析了大型语言模型（LLM）的局限性，指出这些模型虽然在某些任务上展现出惊人的能力，但它们的理解和决策能力仍存在显著的限制。他强调，人们不应被“大模型幻觉”所迷惑，误以为AI具有与人类相似的智能和判断力。
在谈到通用人工智能（AGI）的挑战时，LeCun 强调了当前AI研究面临的困难，包括理解复杂环境、适应新情况以及拥有道德和伦理判断。他认为，AI 的发展需要在解决这些挑战上取得重大突破，而目前的技术远远未达到可能导致人类毁灭的程度。
这场深入的对话为公众提供了一个理解AI的全新视角，提醒人们在欣赏AI技术进步的同时，也要理性看待其局限性和潜在风险。LeCun 的观点提醒我们，尽管AI的发展日新月异，但人类的智慧和控制力在当前阶段仍然不可或缺。来源：AppSo。
英语如下：
News Title: “Turing Award Winner Yann LeCun: AI poses zero risk to humanity, in-depth interview sheds new light on AI perspective”
Keywords: AI safety, Yann LeCun, Chief Scientist at Meta
News Content: In a recent extensive interview with tech blogger Lex Fridman, Yann LeCun, Chief Scientist at Meta and recipient of the Turing Award, expressed his unwavering belief in the safety of artificial intelligence (AI). During the nearly three-hour conversation, LeCun categorically stated that he considers the likelihood of AI dooming humanity to be zero, countering prevalent doomsday scenarios.
LeCun delved into the limitations of large language models (LLMs), explaining that while these models demonstrate astonishing capabilities in specific tasks, they still exhibit significant limitations in understanding and decision-making. He cautioned against the “big model illusion,” warning against the assumption that AI possesses similar intelligence and judgment to humans.
Addressing the challenges of artificial general intelligence (AGI), LeCun highlighted the difficulties currently faced by AI research, including understanding complex environments, adapting to new situations, and having moral and ethical judgment. He argued that significant breakthroughs are required in overcoming these challenges before AI could potentially reach a destructive level, which current technology has not yet attained.
This in-depth dialogue offers the public a fresh perspective on AI, reminding them to appreciate the advancements while acknowledging its limitations and potential risks. LeCun’s insights emphasize that, despite the rapid progress in AI, human wisdom and control remain crucial at this stage. Source: AppSo.
【来源】https://mp.weixin.qq.com/s/ny6awp7BMvnRCZbn4lLTsA
Views: 5