AI Research Factory Completes 135 Scientific Tasks Without Human Intervention
Shanghai's "Golab" AI super research factory completed 135 real scientific research tasks in five days without human intervention, closing the entire research process loop.
The Shanghai Institute for Scientific Intelligence has developed and tested an AI-driven research facility named "Golab" that autonomously completed 135 scientific research tasks over five days without external human assistance. The facility is designed to handle the entire research pipeline, from data analysis to conducting physical experiments.
The "Golab" factory, operated in conjunction with Shanghai Grid Insight Technology Co., Ltd., utilizes a large AI model called "Sui Ren." This model was trained on over 100 million computational data points based on quantum mechanics and statistical thermodynamics, enabling the AI to predict over 50 thermodynamic properties at an industry-leading level.
During the tests, the AI performed tasks ranging from drug discovery and catalyst design to materials research. The system generated molecules, conducted virtual screenings, and sent results directly to an autonomous laboratory for integrated experiments. Verified data confirmed the system's efficiency; for instance, in one catalyst optimization task, activity increased by approximately 15% in the first round and was six times higher than published methods.
This development is part of Shanghai's broader push to advance scientific intelligence. The city has launched the "Science Intelligence Hundred Teams, Hundred Projects" initiative and established a scientific intelligence fund to support young scientists and promote the commercialization of research findings.