陈天桥宣布10亿美元算力支持发现式智能

Core Insights - The first "AI-Driven Scientific Symposium" was held in San Francisco, featuring Nobel laureates and industry leaders discussing how AI can drive scientific discovery [1][2] - Chen Tianqiao announced a $1 billion investment in computational power to support global scientists in "discovery-driven intelligence" research [1] - The symposium highlighted the importance of AI's role in constructing verifiable world models and enhancing human capabilities rather than replacing them [1] Group 1: AI in Scientific Research - Chen Tianqiao emphasized the need for "discovery-driven intelligence" to possess five key capabilities: neural dynamic structure, long-term memory, causal reasoning mechanisms, world models, and metacognitive systems [1] - Omar Yaghi showcased AI's application in materials science, demonstrating a portable device that extracts water from the atmosphere in low humidity conditions using ChatGPT for molecular optimization [1][2] - David Baker presented the RFDiffusion3 model, which enables reverse design of proteins, providing new pathways for research on diseases like Alzheimer's [2] Group 2: AI and Genetic Research - Jennifer Doudna discussed the integration of AI with CRISPR technology, highlighting its potential to enhance understanding of unknown gene functions and advance personalized gene therapy [2] - The symposium concluded with the "AI-Driven Science Prize," recognizing young scientists for their cutting-edge research, indicating a shift towards AI-driven paradigms across multiple disciplines [3] Group 3: Societal Implications of AI - John Hennessy reflected on the rapid adoption of AI, stressing the need for humans to retain key decision-making authority and ensure transparency in AI-generated content [2] - He warned about the potential depletion of global data for AI training in the coming years, noting that improvements in computational energy efficiency have not kept pace with growth [2]