
AI-powered robot perfectly performs surgery after being trained by videos
For the first time ever, a robot has performed a realistic surgery without human help.
According to a new study, the AI-powered robot precisely executed a complex part of a gallbladder removal after being trained by videos of human surgeons. Led by researchers from Johns Hopkins University in Baltimore, the study found that the robot's performance was 'comparable to an expert surgeon' during trials on pig gallbladders.
'This advancement moves us from robots that can execute specific surgical tasks to robots that truly understand surgical procedures,' corresponding author and medical roboticist Axel Krieger said in a news release. 'This is a critical distinction that brings us significantly closer to clinically viable autonomous surgical systems that can work in the messy, unpredictable reality of actual patient care.'
While Krieger and colleagues have trained robots to perform specific surgical tasks in the past, this was the first time that one was able to operate autonomously without a predetermined plan. Rather than teaching a robot to follow a mapped route, Krieger likened their latest breakthrough to 'teaching a robot to navigate any road, in any condition, responding intelligently to whatever it encounters.'
The robot was also able to learn from and respond to voice commands from the team, much like a novice surgeon working under supervision of a mentor. After being trained, the robot operated autonomously on eight pig gallbladders, performing perfectly without any human intervention.
'To me it really shows that it's possible to perform complex surgical procedures autonomously,' Krieger, an associate professor in mechanical engineering, said. 'This is a proof of concept that it's possible and this imitation learning framework can automate such complex procedure with such a high degree of robustness.'
'Comparable to an expert surgeon'
Although the robot took longer than a human surgeon, researchers say it had a 100 per cent success rate in all eight surgeries and was able to adapt to different pig gallbladders and unexpected scenarios, like when its starting position was moved. On average it took the robot five minutes and 17 seconds to complete the required tasks, compared to about four minutes for a human surgeon.
'In these experiments, we observe the consistent ability of our framework to generalize to unseen anatomies and self-correct its own behaviour,' an advance copy of the study explained. 'Preliminary comparison shows that, while our framework requires more operation time, the performance is comparable to an expert surgeon.'
Using the typically human-operated robotic da Vinci Surgical System, the robot was built and trained with the same machine learning architecture that drives ChatGPT.
'Our work shows that AI models can be made reliable enough for surgical autonomy – something that once felt far-off but is now demonstrably viable,' lead author and former Johns Hopkins postdoctoral researcher Ji Woong 'Brian' Kim said in the news release.
Warning: some readers may find the following image disturbing.
Surgical Robot
The surgical robot performing a gallbladder surgery. (Juo-Tung Chen/Johns Hopkins University)
The study was published Wednesday in the journal Science Robotics.
The gallbladder removal procedure required a string of 17 tasks that included identifying certain ducts and arteries in order to place clips and sever parts with scissors. The researchers would now like to train and test the surgical robot for other types of procedures and to eventually perform a completely autonomous surgery.
'Autonomous surgery holds immense potential for improving surgical outcomes, reducing costs, and expanding access to high-quality healthcare,' the study said. '[Our] approach also supports real-time language interventions from expert surgeons, making it practical for potential integration into hospitals as a tool for surgeons to reduce fatigue on simple procedures or for areas with no access to trained surgeons.'
Hashtags

Try Our AI Features
Explore what Daily8 AI can do for you:
Comments
No comments yet...
Related Articles


Globe and Mail
4 hours ago
- Globe and Mail
Oklo Stock: Up 900%, Can It Still Make You Rich?
Oklo (NYSE: OKLO) stock is up 900% -- without a single reactor running. In this video, I'll explore the explosive rise of this nuclear innovator, why it's capturing AI-fueled investor interest, and whether this speculation is genius -- or insanity. *Stock prices used were the market prices of July 5, 2025. The video was published on July 11, 2025. Where to invest $1,000 right now? Our analyst team just revealed what they believe are the 10 best stocks to buy right now. Continue » Should you invest $1,000 in Oklo right now? Before you buy stock in Oklo, consider this: The Motley Fool Stock Advisor analyst team just identified what they believe are the 10 best stocks for investors to buy now… and Oklo wasn't one of them. The 10 stocks that made the cut could produce monster returns in the coming years. Consider when Netflix made this list on December 17, 2004... if you invested $1,000 at the time of our recommendation, you'd have $674,432!* Or when Nvidia made this list on April 15, 2005... if you invested $1,000 at the time of our recommendation, you'd have $1,005,854!* Now, it's worth noting Stock Advisor 's total average return is1,049% — a market-crushing outperformance compared to180%for the S&P 500. Don't miss out on the latest top 10 list, available when you join Stock Advisor. See the 10 stocks » *Stock Advisor returns as of July 7, 2025 Rick Orford has no position in any of the stocks mentioned. The Motley Fool has no position in any of the stocks mentioned. The Motley Fool has a disclosure policy.


Globe and Mail
5 hours ago
- Globe and Mail
PicLumen Launches FLUX.1 Kontext: Prompt-Based AI Image Editing Arrived
PicLumen Integrates FLUX.1 Kontext: Prompt-Based AI Image Editing Arrived Redefining image editing with natural-language control, smarter object logic, and seamless visual transformation. PicLumen, the next-generation AI image creation platform, is proud to announce the launch of its newest capability: FLUX.1 Kontext, a cutting-edge, prompt-based AI image editing model developed by Black Forest Lab. With this powerful new tool, PicLumen users can now edit AI-generated images using plain text commands—no manual retouching, layering, or masking required. From character consistency to background replacement, Kontext offers creators unmatched control and efficiency, reshaping what's possible with one-line instructions. Edit with Prompts. Nothing Else Needed. FLUX.1 Kontext transforms the image editing process into a simple, intuitive dialogue: tell the AI what you want to change, and it will deliver a refined image in seconds. Whether it's replacing an object, fixing distorted text, changing outfits, or correcting details, Kontext understands your intent and re-renders the image accordingly. Core capabilities include: Smart Object Logic: It analyzes image elements and their relationships to maintain realism and visual coherence. Pixel-Level Targeting: Easily identify and edit specific areas like 'the red cup on the left' or 'only change the jacket color.' Creative Style Conversion: Change an image's entire style—Picasso, Van Gogh, Studio Ghibli, and beyond—with a single instruction. Background Control: Remove or swap backgrounds effortlessly using just a prompt. Text Fixing: Clean up distorted or misspelled AI-generated text instantly. Consistency, Context, and Concept Art—All in One Editing Designed to support creative professionals, PicLumen integrates Kontext, allowing you to maintain character identity across different poses, emotions, and angles—ideal for comic creation, concept art, branding, or VTuber development. With its deep understanding of spatial consistency and visual logic, it can generate images that stay true to original designs even when the expression or position changes. Built for Visual Designers, Writers, and Creators Also, this editing tool helps accelerate visual ideation, product mockups, and storytelling. Whether you're uploading sketches for refinement or generating multiple product variants, this tool delivers fast, high-resolution, and visually coherent outputs with minimal effort. PicLumen also supports users with tips and prompt templates for maximum editing accuracy, including best practices like: Using foreground/midground/background structure Locking elements with 'keep original' Iterative re-prompting for subtle refinements How It Works 1. Select an Image: Choose any image you generated in PicLumen. 2. Enter Your Prompt: Describe what you'd like to change—Kontext interprets the instruction. 3. Download Your Result: The refined image appears instantly, ready for use. FLUX.1 Kontext currently supports only images generated within PicLumen. External uploads are not yet supported. About PicLumen PicLumen is a leading AI image generation platform dedicated to making visual creativity more accessible. Through the integration of advanced AI models, intuitive design tools, and a global creator community, PicLumen offers a space where imagination meets innovation. Contact Emily Smith Email: service@ CHENGDU Yile Tech Co., Ltd. Chengdu, Sichuan, China Media Contact Company Name: CHENGDU Yile Tech Co., Ltd. Contact Person: Emily Smith Email: Send Email Country: China Website:


Globe and Mail
5 hours ago
- Globe and Mail
Math Magic Launches World's First 1536^3 Ultra-High-Resolution AI 3D Model Generator Hitem3D.
Hong Kong - Math Magic has officially launched Hitem3D ( the world's first AI-powered tool for 3D content creation that supports ultra-high-resolution models up to 1536^3. Users can upload a single image to instantly generate detailed 3D models, complete with texture support and downloads in multiple file formats. By comparison, most similar tools worldwide currently offer a maximum resolution of only 1024^3. Thanks to Math Magic's proprietary algorithms and technical breakthroughs, Hitem3D can deliver up to 1536^3 resolution. 'This seemingly small difference is actually like upgrading from a GIF to an HD video,' commented an industry researcher. According to Bo Zhang, product lead at Hitem3D, the tool helps designers without 3D modeling skills quickly turn creative ideas into high-precision models that meet production requirements. 'Complex models like mecha, which used to take a senior modeler four to five working days, can now be generated by Hitem3D in under three minutes with hand-crafted detail,' Zhang said. Future plans include adding retopology, multi-view generation, and 3D part segmentation to further reduce creation barriers and improve efficiency. Previously, similar tools could only generate lower-resolution models because existing 3D generation techniques lost critical data during processing, leaving rough shapes and thin details. Users then had to spend significant time refining models to restore detail. Hitem3D's ultra-high-resolution output brings 'what you see is what you get' to AI 3D creation. 'We pioneered direct 1536^3 resolution generation from an image, allowing unprecedented restoration of fine and micro structures. For example, if you upload a macro photo of an ant, Hitem3D can even capture the tiny bristles on its legs,' Zhang said. Math Magic designed a unique sparse convolutional network and fully symmetric 3D VAE architecture, eliminating potential modal conversion errors. Compared to conventional models, reconstruction error (Chamfer Distance) is reduced by 40%. Sparse computation lowers FLOPs by 50%, cutting training and inference costs while ensuring high-quality latent space representation for production models. On platforms like YouTube and X, many 3D creators shared positive reviews, prasing its high accuracy and fine structural reproduction. Several 3D artists reported over 90% efficiency improvements in the first modeling stage, giving digital models more realistic micro-level texture and unlocking new productivity across markets such as 3D printing, product and industrial design, game development, and film production. Website: Video: Media Contact Company Name: Math Magic Contact Person: Celine Su Email: Send Email City: HongKong Country: China Website: