西湖大学许田：AI正在颠覆创新药研发进程

Core Viewpoint - AI is revolutionizing the drug development process in the biopharmaceutical industry, significantly enhancing efficiency and effectiveness [10][14]. Group 1: Human Evolution and Technological Revolution - The essence of humanity is rooted in a plant-based diet, with a shift to meat consumption around three million years ago leading to significant evolutionary changes [4]. - Technological revolutions are the core narrative of human development, characterized by a cycle of slow incubation followed by rapid output and eventual stagnation [6][7]. - Historical technological revolutions include the Industrial Revolution, the rise of the automobile, and the advent of the internet, each marking significant economic and societal shifts [7]. Group 2: Current Technological Waves - The world is currently experiencing overlapping technological waves, including the ongoing information network revolution and the emerging biopharmaceutical and AI waves [8]. - AI is reshaping social cognition and information dissemination, leading to changes in societal behavior and decision-making processes [8]. Group 3: Biopharmaceutical Industry Insights - In 2022, China's biopharmaceutical industry achieved over $100 billion in external licensing, surpassing the export value of new energy vehicles, indicating a significant growth trajectory [8]. - Despite lower investment levels compared to the new energy vehicle sector, the biopharmaceutical field's growth reflects the irreversible nature of technological waves [8]. Group 4: AI in Drug Development - The integration of AI in biopharmaceuticals has led to the establishment of successful companies, with a 90% success rate in projects initiated by the speaker's team, resulting in 10 companies being launched [11]. - AI has drastically reduced drug development timelines, with three drugs entering clinical phase II within six years, marking them as "First-in-Class" innovations [14]. - AI algorithms have enabled the standardization of biopharmaceutical data, expanding usable patient data from thousands to billions, thus enhancing the training of AI models [14].