Nature重磅：智能的尽头是算力，谷歌大佬承认「预测下一个词即智能」

Core Insights - The article discusses the shift in understanding AI development, emphasizing that intelligence growth is not solely dependent on chip speed but rather on structural reorganization and collaboration among multiple units [1][7][16] Group 1: AI Development and Structure - The traditional view of Moore's Law, which posits that faster chips lead to stronger intelligence, has been challenged as chip speeds plateaued around 2020 [1][7] - Despite the stagnation in chip speed, AI has continued to evolve rapidly, with large models demonstrating unprecedented capabilities [7][8] - The article posits that the enhancement of intelligence is achieved through collective participation in prediction rather than individual acceleration [6][8] Group 2: Collective Intelligence and Collaboration - The concept of collective intelligence is highlighted, where groups, rather than individuals, enhance predictive capabilities through collaboration [6][9] - Modern AI development mirrors natural intelligence evolution, relying on the parallel processing of numerous simple computational units rather than singular advanced capabilities [8][12] - The article introduces the idea of "technical symbiotic generation," where AI's rise is seen as a natural progression in the history of intelligence [6][8] Group 3: Future of Intelligence - The future of intelligence is framed as a collaborative network involving both humans and machines, rather than a competition between them [12][14] - AI is positioned as an integral part of a larger cognitive system, enhancing decision-making capabilities across various complex structures [9][12] - The article concludes that the emergence of AI is a natural extension of intelligence evolution, emphasizing the importance of structural organization and collaboration in achieving higher levels of capability [16][17]