A team of researchers have used videos to train surgical robots to perform operations on their own with the skill of human doctors. The robots learned to suture wounds, manipulate needles and perform other tasks, and surprisingly, the robots even corrected their own errors, according to the Johns Hopkins University and Stanford University researchers.

The researchers equipped their existing da Vinci Surgical System with a machine-learning model that analyzes surgical procedures recorded by cameras mounted on the robot’s instruments. The videos provide a detailed representation of the tasks performed by skilled surgeons.

The new model goes one step further, says SiliconAngle, by including a feedback mechanism that allows the robot to analyze and even correct its own performance to achieve outstanding levels of precision and dexterity.

According to The Washington Post, robots have been helping doctors during surgery for decades by using joystick-like controllers to manipulate arms and grippers to perform various procedures. Now, scientists say they’ve trained robots to work autonomously. They can manipulate needles, tie knots and suture wounds, as well as fetch dropped instruments.

The next hurdle will be to see how well these skilled robots can perform entire surgeries on animal cadavers. Once the robots are deemed ready, researchers hope they will fill in for the shortage of surgeons in the U.S. medical industry.   

Some estimates predict that by 2036, our aging population combined with a dearth of surgical students will lead to a shortage of between 10,000 and 20,000 surgeons, says the American Association of Medical Colleges. While critics say that robotic surgery is more costly than traditional operations, proponents counter that increased skill levels of the robots and their availability to perform will outweigh the negatives.  

Axel Krieger, an associate professor at Johns Hopkins Whiting School of Engineering, who supervised the research, said, “In our work, we’re not trying to replace the surgeon. We just want to make things easier for the surgeon.”

Critics argue that surgeons need to study imaging of patients before they decided on the correct operating technique. If robots are to perform surgery solo, they will have to understand all the imaging revealed in MRIs and CT scans. Another obstacle is how medical mishaps, such as bleeding or improperly placed sutures, will be handled.

“If a blunder occurs, who hold the responsibility?” asks Dr. Amer Zureikat, the director of robotic surgery at the University of Pittsburgh Medical Center. “Is it the doctor? Is it the AI developer? Is it the hospital facility? Is it the robot manufacturer?”

Privacy is another obstacle. The autonomous robots were not trained using actual videos of surgeries, but they will need that exposure if they are to operate solely on their own. That means patients will have to give their permission to have their surgeries shared. Zureikat raises other questions about which patients will have access to the new technology and whether skilled surgical robots will make human surgeons outdated and inept in the long run.