Nature重磅：“AI科学家”真的来了，自主开会搞研究，几天时间设计出抗病毒纳米抗体

Core Viewpoint - The article discusses the development of an AI-driven virtual laboratory that enables multidisciplinary research teams to tackle complex scientific problems, specifically in the context of designing new nanobodies for SARS-CoV-2 [2][4][11]. Group 1: AI-Driven Virtual Laboratory - Researchers from Stanford University and the Chan Zuckerberg Biohub have created a virtual laboratory platform powered by AI agents, which can autonomously design and execute complex research strategies [2][4]. - The virtual lab allows human scientists to pose scientific questions while AI agents, including a "Chief Scientist Agent" and various "Specialist Agents," collaborate to advance research [5][6]. Group 2: Research Outcomes - Within days, the virtual laboratory successfully designed 92 novel nanobodies, with two showing the ability to bind to the spike protein of new SARS-CoV-2 variants during laboratory validation [9][11]. - The research demonstrates that AI agents can generate creative and rational solutions to scientific challenges, enhancing human scientists' capabilities rather than replacing them [11]. Group 3: Implications and Future Applications - This study marks the first instance of autonomous AI agents effectively solving a challenging scientific research problem from start to finish, showcasing a new paradigm where AI drives the entire research process [11]. - The virtual laboratory platform is designed for biomedical research but can be adapted for broader scientific fields, indicating its potential for widespread application [11].