AI Robot Performs Surgery
Recently, a robot performed part of a gallbladder removal surgery- all by itself.
A research team at Johns Hopkins University created and trained a robot to perform part of a gallbladder removal surgery by itself! The surgery contains three main steps:
Opening the abdomen to locate the gallbladder
Cutting the blood vessels holding the gallbladder in place
Removal of the organ
These researchers taught the robot to perform step two on a pig gallbladder (we aren’t at the level of human surgeries yet!), which involves identifying the vessels, clipping them to prevent too much blood loss, and cutting them to free the organ.
This is not the first robot to do surgery- surgical robots have been around to automate surgical tasks for years- the first one was created in 1985, and they became officially FDA-approved in 2000. However, what makes this new robot different is that it’s powered by artificial intelligence. This means instead of repeatedly following the same pre-programmed steps like its predecessors, it can respond to voice commands from a surgical team, adapt to unexpected situations, and self-correct when things go off course!
Training surgical robots
AI has changed the game in medical robotics. The artificial intelligence network (called SRT-H) that this new robot uses was created in a way that mimics how the human brain works- similar to ChatGPT. It learns through experience and training, and tunes its surgical actions based on the information it’s fed.
To train it, researchers showed it 17 hours worth of videos of actual gallbladder surgeries, paired with text captions describing each individual surgical step. That way, it was able to see the movements it needs to make while using the captions to gain context about why each step of surgery is important for the goal of removing the gallbladder. The goal was for the robot to develop a true understanding of this procedure.
How did it do?
This AI robot learned to identify blood vessels, position the clips, and make the incisions necessary to remove the organ with 100% accuracy!!!
Albeit, it moves slower than human surgeons (taking ~5 minutes for the entire part of the procedure), but is still just as accurate. Fascinatingly, it was able to respond to real-time spoken instructions from human surgeons during the procedure, helping it learn to recognize when it made incorrect movements and eventually correct its mistakes without human intervention..almost like a real (human) surgical trainee!
Even when they started the robot from awkward angles and poses, it was still successfully able to grasp the gallbladder. Sometimes it would miss its target- but it would realize this BEFORE making any permanent cuts, readjusting its movements to perform the correct cuts.
Should we be nervous about this?
I’d say a little…
While the idea of automating some surgeries with AI is exciting, there are reasons to pause. These AI systems aren’t always predictable. They rely on vast amounts of data storage which has high environmental costs, and if done in humans could raise uncomfortable questions about accountability and consent.
But here’s the other side of that coin: the medical system as it currently exists isn’t perfect either. Surgeons often work dangerously long hours with limited rest. If I had to choose between a robot trained on hundreds of surgical procedures who does it right 100% of the time and a human who’s been awake for 20 hours straight… I’d be torn.
Still, this isn’t an all-or-nothing debate. A robot is not the best solution to overworked physicians. We could just train and hire more human surgeons. Expand residency programs. Improve work-life balance. Address systemic issues without handing it all over to machines.
That said, this advancement in surgical robotics is undeniably huge. It's exciting- but concerning. And it raises deeper questions we should all be asking, about what AI can do, and what it should do.
Would you trust a robot to operate on you?
References
Kim, J. W. B., Chen, J. T., Hansen, P., Shi, L. X., Goldenberg, A., Schmidgall, S., Scheikl, P. M., Deguet, A., White, B. M., Tsai, R., Cha, R. J., Jopling, J., Finn, C., & Krieger, A. (2025). SRT-H: A hierarchical framework for autonomous surgery via language-conditioned imitation learning. Science robotics, 10(104), eadt5254. https://doi.org/10.1126/scirobotics.adt5254
Robot surgery through the years: https://thesurgicalclinics.com/history-of-robot-assisted-surgery/
Johns Hopkins news article: https://hub.jhu.edu/2025/07/09/robot-performs-first-realistic-surgery-without-human-help/
These of course are near perfect conditions, it will be interesting to see how well AI can cope with the abnormal circumstances which normally make such surgery necessary.
Also, the mistakes AI might make will likely be different than the ones humans would. Will we be willing to accept that gramma died under the AI knife because of something a human never would have allowed to happen, because the statistical data still shows AI is overall safer?