Researchers at Johns Hopkins University say an artificial-intelligence system has autonomously completed a realistic gallbladder removal with 100% accuracy, marking a significant advance toward fully robotic surgery. Details of the work, led by mechanical-engineering associate professor Axel Krieger, were published Wednesday in the peer-reviewed journal Science Robotics. The system, called Surgical Robot Transformer-Hierarchy (SRT-H), was trained on roughly 17 hours of annotated surgical video—about 16,000 individual instrument motions—and then performed eight ex-vivo cholecystectomies on human-like tissue models. During each 17-step operation the robot identified ducts and arteries, placed clips, and cut tissue, adapting when dye altered organ appearance. It also responded to spoken commands such as “grab the gallbladder head,” learning from the feedback much like a junior surgeon. While the procedure took longer than with an experienced human, researchers said the technical quality was comparable. Krieger described the achievement as a shift “from robots that can execute specific tasks to robots that truly understand surgical procedures.” The team, which includes former Johns Hopkins post-doctoral fellow Ji Woong Kim now at Stanford University, plans to extend the approach to other operations and ultimately to human trials. External specialists welcomed the milestone but cautioned that rigorous testing and regulatory review will be required before autonomous robots work on live patients.