苦涩教训提出者再发声：LLM扩展不能只靠算力
原标题：The inventor of the bitter lesson has a bitter lesson for LLM scaling enthusiasts
AI 摘要
图灵奖得主Richard Sutton发表新观点，认为仅靠算力堆叠不足以实现智能，AI科学需要更聪明的新方法，且远未完成。他此前已批评单纯扩展大语言模型的局限，称合成数据是重大错误，并呼吁构建具备世界模型、能从自身经验持续学习的智能体。这一观点在AI社区引发广泛讨论，Jerry Tworek等研究者呼应称算法重要性随算力增长而提升，而Sam Altman和Dario Amodei等产业领袖则坚持LLM扩展仍有巨大空间。
正文节选
The inventor of the bitter lesson has a bitter lesson for LLM scaling enthusiasts Key Points - Turing Award winner Richard Sutton, author of the influential 2019 essay "The Bitter Lesson," writes in a new post that raw compute alone isn't enough for intelligence. AI science needs clever new methods and is "far from finished." - Sutton had already criticized the limits of simply scaling large language models, called synthetic data a "big mistake," and called for agents with world models that cont