在周二于Meta伦敦举办的一次活动中，图灵奖得主、Meta AI负责人Yann LeCun重申了他对通用人工智能（AGI）的怀疑立场。LeCun认为，人类智能并不具备普遍性，因此“通用人工智能”这一概念本身并不现实。他倾向于追求更为实际的“人类水平的AI”。
在讨论中，LeCun指出现有的AI系统在四个关键的认知能力上存在缺陷：推理、规划、持久记忆和理解物理世界。这些不足导致AI在实际应用中可能受限且错误频发，例如自动驾驶汽车的安全问题、家用机器人的家务处理能力，以及智能助手仅能完成基础任务的局限性。
LeCun特别批评了大型语言模型（LLMs），尽管这些模型在语言流畅性上表现出色，但它们对现实世界的理解极其有限。他强调，人类通过与环境的交互来学习，而不仅仅是通过阅读文本。他估计，一个四岁的孩子在成长过程中接触到的数据量可能是最大LLMs的50倍，这进一步凸显了当前AI模型的局限性。
LeCun的观点为AI研究领域提供了重要的反思点，暗示了在追求人工智能的道路上，模拟人类与环境的交互以及增强系统在复杂情境中的理解能力将至关重要。
英语如下：
**News Title:** “Turing Award Winner Yann LeCun Warns: Large Models Fall Short of Human Intelligence, AI Faces Four Major Cognitive Challenges”
**Keywords:** Turing Award, Large Models, Human Intelligence
**News Content:**
### Yann LeCun, Turing Award Winner, Casts Doubt on Large Models’ Ability to Reach Human-Level Intelligence
During an event in London hosted by Meta on Tuesday, Yann LeCun, a Turing Award recipient and head of Meta AI, reaffirmed his skepticism about the concept of artificial general intelligence (AGI). LeCun argues that human intelligence is not universal, making the idea of “general AI” unrealistic. Instead, he advocates for pursuing more practical “human-level AI.”
In the discussion, LeCun pointed out shortcomings in four key cognitive abilities of existing AI systems: reasoning, planning, long-term memory, and understanding the physical world. These deficiencies limit AI’s practical applications and can lead to errors, such as safety issues with autonomous vehicles, the limitations of household robots, and the basic task capabilities of AI assistants.
LeCun specifically criticized large language models (LLMs), acknowledging their fluency in language but highlighting their limited understanding of the real world. He emphasized that humans learn through interaction with their environment, not just by reading text. He estimates that a four-year-old child is exposed to 50 times more data in their development than the largest LLMs, underscoring the limitations of current AI models.
LeCun’s perspective offers important food for thought in the AI research community, suggesting that emulating human interaction with the environment and enhancing systems’ understanding in complex situations will be crucial in the pursuit of artificial intelligence.
【来源】https://thenextweb.com/news/meta-yann-lecun-ai-behind-human-intelligence
Views: 2