Search strategy and selection criteria

This narrative perspective draws on publications indexed in MEDLINE between January 2015 and May 2025 that examine AI-driven decision-making, AR-guided navigation, semi-autonomous robotics, and real-time imaging in surgical settings. All search results were exported, de-duplicated, and curated in Mendeley Desktop 1.19.8, which served as the reference-management tool throughout the review process.

Augmented reality in operative visualization

AR is revolutionizing surgical practice by elevating visualization, spatial orientation, and intraoperative decision-making [7]. By overlaying digital information, such as 3D anatomical models, surgical trajectories, and real-time imaging, directly onto the surgeon’s field of view. AR provides immersive, real-time guidance during complex procedures, potentially including the precise administration of localized therapeutic agents. AR-based navigation systems, for example, have demonstrated sub-millimeter accuracy in projecting critical structures onto the surgical field, improving spatial orientation and reducing the risk of iatrogenic injury [7]. Moreover, AR has been shown to enhance surgeon confidence and precision during delicate operations by providing virtual navigation aids that align with the patient’s anatomy [8]. AR platforms may allow surgeons to visualize and interact with holographic representations of patient-specific anatomy, facilitating better planning and execution of surgical interventions, and envisioning future applications in guiding targeted drug delivery [9].

Real-time imaging for surgery

Real-time imaging technologies provide dynamic, intraoperative visualization of anatomical structures during procedures. Tools such as intraoperative ultrasound, real-time magnetic resonance imaging (MRI), and fluorescence-guided imaging are increasingly used to delineate tumor margins, assess vascular integrity, and adapt surgical strategies in real time [10]. For example, indocyanine green (ICG) fluorescence imaging is widely applied in hepatobiliary and colorectal surgeries to assess tissue perfusion and detect bile leaks or metastatic lymph nodes with high sensitivity [11]. In neurosurgery, intraoperative MRI enables real-time visualization of brain shifts during tumor resection, allowing surgeons to update navigation systems and improve resection accuracy [12]. To enable robotic assistance in surgery, MRI-compatible platforms are constructed from non-ferromagnetic composites and rely on pneumatic or piezoelectric actuation with fiber-optic feedback, thereby preserving sub-millimetric precision without distorting the magnetic field [13]. Advanced imaging platforms integrated with AI algorithms can highlight pathological tissue in real time based on texture, color, or fluorescence signals, supporting more precise excision and reducing operative time [14].

AI-assisted decision support

AI has become an essential element in the evolution of precision surgery, impacting every stage of the surgical process, from preoperative planning to intraoperative execution and postoperative management. By utilizing its ability to process large volumes of complex data, AI supports a more integrated and personalized approach to surgical care, in line with the core goals of precision medicine.

Clinical applications of AI are rapidly expanding across the surgical pathway (Figure 2). In the diagnostic phase, AI-assisted colonoscopy systems have significantly improved the detection of adenomas and polyps in colorectal cancer, surpassing conventional methods and enabling earlier therapeutic interventions [15]. AI further enhances diagnostic accuracy by combining radiological and histopathological data to evaluate tumor aggressiveness. In preoperative planning, AI tools used in orthopedic surgery improve implant positioning in hip and knee arthroplasty [16], while in hepatic and neurosurgical procedures, AI-driven 3D segmentation allows for precise identification of anatomical landmarks and surgical margins. Intraoperatively, systems are integrated with robotic platforms and AR interfaces to support real-time visualization of subsurface anatomy and guide precise surgical actions [17]. Additionally, “surgical black box” platforms allow AI to analyze intraoperative video, audio, and device data, identifying performance deviations and improving surgical safety [18]. AI-driven imaging analysis also aids in distinguishing between healthy and pathological tissue during resection, while in endoscopy, it significantly increases polyp detection through automated visual alerts [15, 18]. Postoperatively, machine learning models that incorporate clinical, demographic, pharmacogenomic, and laboratory data have demonstrated superior performance in predicting complications such as infection, hemorrhage, and readmission. Furthermore, these models can optimize personalized drug regimens, predict adverse drug reactions, and tailor analgesic strategies, enabling more responsive and personalized perioperative care strategies [19].

Display full size

Figure 2. The historical timeline of augmented reality, real-time imaging, artificial intelligence (AI), and robotics.

Robotic surgery: A ChatGPT-like revolution in robotics is imminent

Over the past decade, minimally invasive surgery has undergone a rapid evolution with the advent of robotic-assisted techniques (Figure 2). The da Vinci platform emerged as a pioneering system and, since its approval by the Food and Drug Administration (FDA), has been widely adopted across multiple surgical specialties [20]. Robotic surgery has been associated with improved clinical outcomes, including reduced intraoperative blood loss and transfusion requirements, shorter hospital stays, and a lower rate of complications compared to traditional open or laparoscopic approaches [20]. Subsequently, other multiport robotic surgical systems have been developed, highlighting the ongoing expansion and innovation in this field [21].

Very recently, robotic surgery marked a new historic milestone when a thoracic surgeon performed a lung cancer operation from a robotic console in China, while the patient was located over 8,000 kilometers away in Romania [22]. The procedure was conducted using a 5G connection that enabled near real-time synchronization between the surgeon’s commands and the movements of an uniportal robotic arm. Through a single incision, the right upper lobe of the lung was successfully resected. Remarkably, the latency across this intercontinental connection was only 0.125 seconds, almost imperceptible [22].

It is frequent that clinicians and surgeons lag behind the latest developments in technology, including robotics; any innovation must first undergo extensive medical, ethical, and legal evaluation before it can be safely implemented in clinical practice. To understand where robotic surgery is heading, it is essential to look beyond the operating room and examine the broader landscape of robotics, where emerging trends often signal the future direction of surgical innovation.

One of the most recent breakthroughs is the potential integration of AI into robotics. Traditionally, robots have been trained to perform specific tasks through direct human instruction. For instance, a robot could be programmed to transfer an object from one hand to another, but it would not be able to respond to a greeting unless explicitly coded to do so. This paradigm has changed thanks to AI systems by enabling robots to adapt autonomously to new situations, objects, and environments without the need for prior task-specific training. This is possible through the integration of natural language understanding, visual perception, and the execution of precise physical actions [23]. This concept has been further advanced by introducing the notion of a visuo-linguistic actuator (VLA) as a core component of its system [24]. A VLA is an architecture that combines three key capabilities: vision (visuo), language (linguistic), and physical action (actuator). The goal is to create a unified AI model that enables robots to interact with the world in a more natural, generalist, and context-aware manner (Figure 3). Their latest model has been trained to understand and follow natural language instructions (e.g., “move the light”), process visual information from the environment (via cameras or sensors) and translate these multimodal inputs into complex physical actions (such as moving a robotic arm or grasping objects), hinting at future capabilities for precise, AI-guided surgery, including administration of intra-surgery drugs [24].

Display full size

Figure 3. How robots interact with the world by using VLA (visuo-linguistic actuator) approach.

In essence, these systems go beyond executing pre-programmed commands; they are capable of inferring about what they see and hear, and then deciding how to act accordingly. This represents a significant step toward more autonomous and intelligent robotic systems that could reshape the way surgical robotics is practiced. In this sense, AI is rapidly becoming an integral component of robotic surgery, enhancing the surgeon’s capabilities through data-driven insights and emerging autonomous functionalities. One of the most promising developments is the incorporation of computer vision and AI algorithms to provide real-time intraoperative guidance [5]. For example, by showing suggestions, alerts, or the delineation of anatomical structures identified automatically by AI systems as the surgeon operates. These recommendations are expected to become significantly more accurate and context-aware in the coming years. This digital “second look” can highlight potential risk zones or guide dissection planes, acting as a silent safety assistant throughout the procedure [25].

Concerning the level of automation, although the surgeon retains primary control, various functions are being developed to delegate repetitive or high-precision tasks to the robotic system. AI may soon enable robotic platforms to perform simple surgical actions semi-autonomously, such as incision closure or knot tying [6]. Prototype systems already exist in which the surgeon positions the instruments and, with a single command, the robot completes the suture autonomously. Notably, in 2022, researchers at Johns Hopkins University achieved a significant milestone by performing the first fully automated laparoscopic surgery, where an autonomous robotic system independently executed a bowel anastomosis in an animal model without human intervention [26]. Another innovative application of AI lies in performance analysis and surgical training. Newer platforms are incorporating real-time analytics and collaborative tools. These systems can collect objective data from each procedure, including instrument movement, applied forces, and task efficiency, and provide personalized feedback to the surgeon, promoting continuous technical improvement [27].

The future of robotic surgery is expected to be shaped by major advances. The fifth-generation da Vinci system aims to incorporate a computing system that is significantly more powerful than its predecessors, offering up to 10,000 times the processing capacity of the previous model [27]. This leap in performance is expected to enable future upgrades that may be retrofitted into existing da Vinci platforms, further extending their capabilities [27]. Although still in its early stages, this convergence of AI and robotic systems points toward a future in which automation and remote surgery could significantly expand access to high-quality surgical care, while maintaining rigorous standards of precision and patient safety.