**正文：**
近日，在Meta伦敦的一次活动中，图灵奖得主、Meta AI负责人Yann LeCun表达了对通用人工智能（AGI）的深刻怀疑。LeCun认为，所谓的“通用人工智能”并不存在，因为人类智能本身并不具备普遍性。他更倾向于追求达到“人类水平的人工智能”。
LeCun指出，当前的AI系统还面临着四个基本的认知挑战：推理、规划、持久记忆和理解物理世界。这四大挑战使得AI应用在实际场景中的应用受限，并且容易出错。例如，自动驾驶汽车在公共道路上的安全性问题、家用机器人在处理基本家务方面的困难，以及智能助手只能完成基本任务，这些都反映了AI目前的局限性。
特别是对于大型语言模型（LLMs），LeCun特别强调它们的局限性。虽然LLMs在语言流畅性方面表现出色，但它们对现实世界的理解非常浅薄。他认为，人类通过与世界的互动来获取信息，而不是仅仅通过阅读文本。他甚至估计，一个四岁的孩子所接触到的数据量可能是世界最大LLMs的50倍。
LeCun的这番言论为我们揭示了AI的局限性，也让我们重新思考什么是真正的人工智能。在AI的发展道路上，我们还有很长的路要走。
英语如下：
### News Title: **Yann LeCun, Head of Meta AI, Denies the Existence of General Artificial Intelligence**
Keywords: Yann LeCun, AI skepticism, Human-level intelligence
#### News Content:
**Title: Turing Award Winner Yann LeCun Questions the Existence of General AI and Emphasizes Its Limitations**
**Body:**
Recently, at a Meta event in London, Yann LeCun, a Turing Award winner and the head of Meta AI, expressed profound skepticism about the existence of General Artificial Intelligence (AGI). LeCun believes that the concept of “general artificial intelligence” is fallacious because human intelligence itself is not universal. He is more inclined to pursue the goal of achieving “human-level artificial intelligence.”
LeCun pointed out that current AI systems are still facing four fundamental cognitive challenges: reasoning, planning, persistent memory, and understanding the physical world. These four challenges limit the application of AI in real-world scenarios and make it prone to errors. For instance, the safety concerns of autonomous cars on public roads, the difficulties faced by home robots in performing basic household tasks, and the limited capabilities of smart assistants in completing fundamental tasks all reflect the current limitations of AI.
LeCun particularly emphasized the limitations of Large Language Models (LLMs). While LLMs excel in linguistic fluency, their understanding of the real world is quite superficial. He believes that humans acquire information through interaction with the world, rather than just by reading text. He even estimates that the amount of data a four-year-old child is exposed to may be 50 times more than that of the world’s largest LLMs.
LeCun’s comments reveal the limitations of AI and prompt us to reconsider what true artificial intelligence really is. On the path of AI development, we still have a long way to go.
【来源】https://thenextweb.com/news/meta-yann-lecun-ai-behind-human-intelligence
Views: 5