Key points
The effective integration of artificial intelligence (AI)-enabled robotic systems into surgical practice requires further improvement in how humans and machines communicate intent, coordinate actions, and operate within defined safety constraints—maintaining a common understanding of the task, environment, and outcomes.
Progress depends on addressing two fundamental questions: how is clinical intent specified to the system and how does the system execute actions safely and reliably?
Hybrid approaches that combine AI models with patient-specific “digital twins” could support real-time understanding, planning, and safe execution in surgical settings.
Introduction
Recent advances in robotic technology, computation, and artificial intelligence (AI) are enabling increasingly powerful human–machine partnerships that will enhance patient care in surgery and other fields of interventional medicine. By exploiting the complementary capabilities of surgeons and robotic systems, these partnerships can provide significant benefits to patients, enabling safer, more precise, and less invasive procedures, and, in some cases, making previously infeasible procedures possible. In addition, these systems inherently function as “flight data recorders”, generating rich procedural data that can be coupled with outcomes using machine learning methods to improve future practice. However, the central challenge is not simply to increase robotic autonomy but also to ensure that humans and machines can communicate intent effectively, operate within well-defined safety constraints, and maintain a shared understanding of the surgical environment.
In their lead article, Granados et al. address the issues arising from the introduction of AI and robotic systems into surgical practice (). One of the main points raised is that this emerging partnership merits broad discussion and consideration of its impact on multiple stakeholders in surgery and interventional medicine. The article includes a discussion of advances in large AI models and robotic systems in relation to surgical workflows, team roles and responsibilities, and other aspects of surgical practice. The authors raise many interesting points, including societal issues (e.g., liability, bias, and equity) associated with the integration of these systems into surgical practice. Here, I comment on several related issues focusing on communication, control, and shared situational awareness in human–robot surgical teams.
Autonomy and communication in robotic surgical assistant systems
The potential of robotic systems to extend human sensorimotor capabilities has long been recognized (). For example, they can enable tremor-free tissue manipulation with extraordinary precision, sense tool-to-tissue interaction forces an order of magnitude below human perceptual thresholds, and permit highly dexterous manipulation inside a patient’s body (). They do not experience fatigue and can operate in infectious environments or in the presence of ionizing radiation. As Granados et al. note, these capabilities have typically been used through some form of teleoperation or other direct human-in-the-loop control.
However, the emergence of increasingly powerful AI is enabling the development of surgical assistant systems that can act with less direct human control. A common trend in the current literature is to categorize surgical robots on a five- or six-point scale, ranging from no autonomy (e.g., a scalpel) to full human-like autonomy (). While this taxonomy provides a useful conceptual framework, it can obscure the more fundamental challenges involved in integrating AI capabilities into surgical systems. Real systems typically exhibit characteristics of multiple levels of autonomy.
The fundamental questions therefore are not how to assign a level of autonomy but rather (i) how the clinical team can tell the robot what it is supposed to do and (ii) how the robot can execute those instructions while ensuring that it will not do what it is not supposed to do.
Viewed in this context, surgical robots have exhibited significant autonomy from the beginning. For example, radiation therapy machines are arguably among the earliest surgical robots, and their operation is highly computer-integrated. A human dosimetrist uses a powerful computer to develop a treatment plan from computed tomography images. In the treatment room, the machine is registered to the patient and delivers beams of high-energy radiation from different angles to produce the desired radiation pattern. Similarly, the ROBODOC system (Integrated Surgical Systems, now Think Robotics) for joint replacement surgery autonomously machines the patient’s femur and tibia to receive an orthopedic implant, while the surgeon monitors the machining operation and performs other parts of the procedure manually (). In both cases, the robot’s ability to perform a complex motion task with high precision is coupled with a human–computer partnership in planning a procedure increasingly supported by machine learning methods [e.g., ()].
The widespread adoption of teleoperated surgical robots for endoscopic surgery has led to considerable academic interest in developing AI-based methods to automate specific surgical subtasks, such as endoscope manipulation, suturing, retraction, and debridement. There has also been recent progress toward substantial automation of a full cholecystectomy (). However, this work is still far from ready for use in human surgery. As Granados et al. note, robotic autonomy will likely be introduced in a more gradual way for tasks that can exploit the capabilities of the robot while still retaining surgeon supervision. Key challenges include safety, generalizability, and unambiguous specification of human intent. One emerging theme is the exploration of the use of augmented reality to integrate human assistants more fully into the surgeon–robot team ().
A significant potential advantage of surgical robots is their ability to enforce safety barriers to prevent unintentional damage to safety-critical structures, even in cases where the surgeon is guiding the surgical tool. One recent example is the work at Johns Hopkins University to protect delicate structures in head-and-neck surgery applications [e.g., ()]. These efforts rely on maintaining an accurate registration between the robotic tools and computer models of the structures to be protected. A key enabler of this work is the development of real-time vision methods to provide direct feedback to the robot, similar to the surgeon’s own visual feedback while manipulating tissue. An important next step may be to combine these capabilities with surgical phase recognition to enhance overall surgical team performance.
Shared situational awareness
The challenges of communication and control ultimately reduce to a deeper issue: how humans and machines represent and share an understanding of the surgical task and environment. Any cooperative endeavor—whether between humans, humans and robots, or robots with each other—requires that the parties involved share a common understanding of the task, the environment, their capabilities, and the results of actions taken. The fundamental issue is representational. As Granados et al. note, one common approach has been the use of vision-language models and vision-action models, in which situational awareness is encoded in convolutional neural network representations. These methods have had notable successes, but they also have limitations. They often lack explainability and can be difficult to integrate with visualization or robot command-and-control methods.
One promising alternative is to combine these network-based methods with more explicit “digital twin” simulation models of individual patients (). Offline uses of these models include procedure planning, training of surgeons, and the generation of data with known ground truth for training neural networks. Intraoperatively, these can be updated using real-time sensing and used to support mixed-reality visualizations, safety constraints, and other enhanced control modes for the robot.
The need for shared situational awareness extends beyond surgeon–robot interaction to the entire surgical team. There is an emerging trend to equip operating rooms with a “black box” that combines vision, audio, and other intraoperative data streams. These systems can be combined with machine learning methods to monitor surgical workflow and may support the development of near-miss warning systems and other clinical applications.
Conclusion and further thoughts
One of the greatest strengths of the Granados et al. article is its broad focus on the entire perioperative environment. Successful surgical interventions depend on the coordinated efforts of the entire surgical team, not just the surgeon. The article’s discussion of ethics, health equity, and regulatory issues raises important considerations, as does its treatment of how AI and robotic technologies may reshape team roles and responsibilities.
The impact of these technologies will extend well beyond the operating room. Surgical interventions should be understood as one element of a larger care continuum that comprises diagnosis, treatment planning, surgical intervention, postoperative care, follow-up, and long-term management. As AI-enabled robotic assistants are integrated into these areas, they have the potential to reshape workflows throughout the healthcare system. Ultimately, progress will depend on developing robust approaches to communication, shared representation, and coordinated decision-making between humans and machines.
Statements
Author contributions
RHT: Conceptualization, Writing – review & editing, Writing – original draft.
Funding
The author declared that financial support was not received for this work and/or its publication.
Conflict of interest
The author declares that this work was conducted in the absence of financial relationships that could be construed as a potential conflict of interest. The author holds approximately 100 US and International patents, some of which may be relevant to the themes of this article and some of which have been licensed by Johns Hopkins University to Intuitive Surgical, Varian Medical Systems, Philips Nuclear Medicine, Virtuoso Technologies, LIV Medical, Oncospace Inc., Galen Robotics, and other corporate entities. The author and the University are entitled to royalty distributions related to this technology, and the author has received or may receive some portion of these royalties. These arrangements have been reviewed and approved by JHU in accordance with its conflict-of-interest policy.
Generative AI statement
The author declared that generative AI was used in the creation of this manuscript. The author used ChatGPT 5.3 to perform light copy-editing for this manuscript.
Publisher’s note
All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.
References
1
GranadosAKhannaRFischerNRaisonNCiabattinMRobertshawHet al. Evolving surgical teams in the age of artificial intelligence and robotics. Front Sci (2026) 4:1783803. doi: 10.3389/fsci.2026.1783803
2
TaylorRH. A perspective on medical robotics. Proc IEEE (2006) 94(9):1652–64. doi: 10.1109/JPROC.2006.880669
3
GuthartGSSalisburyJK. The Intuitive™ Telesurgery System: overview and application. In: Proceedings of the IEEE International Conference on Robotics and Automation (ICRA2000); 2000 Apr 24–28; San Francisco, CA (2000). 618–21. Available at: https://ieeexplore.ieee.org/stamp/stamp.jsp?arnumber=844029
4
YangG-ZCambiasJClearyKDaimlerEDrakeJDupontPEet al. Medical robotics—regulatory, ethical, and legal considerations for increasing levels of autonomy. Sci Robotics (2017) 2(4):eaam8638. doi: 10.1126/scirobotics.aam8638
5
TaylorRHPaulHAKazanzidesPMittelstadtBDHansonWZuharsJFet al. An image-directed robotic system for precise orthopaedic surgery. IEEE Trans Robotics Automation (1994) 10(3):261–75. doi: 10.1109/70.294202
6
EskandarK. The role of artificial intelligence in orthopedic surgery: current applications and future perspectives—a systematic review of the literature [Spanish]. Rev Esp Cir Ortop Traumatol (2025) S1888441525001778. doi: 10.1016/j.recot.2025.09.005
7
KimJWChenJ-THansenPShiLXGoldenbergASchmidgallSet al. SRT-H: a hierarchical framework for autonomous surgery via language-conditioned imitation learning. Sci Robotics (2025) 10(104):eadt5254. doi: 10.1126/scirobotics.adt5254
8
QianLDeguetAKazanzidesP. ARssist: augmented reality on a head-mounted display for the first assistant in robotic surgery. Healthcare Technol Lett (2018) 5(5):194–200. doi: 10.1049/htl.2018.5065
9
IshidaHSahuMMunawarANagururuNGalaiyaDKazanzidesPet al. Haptic-assisted collaborative robot framework for improved situational awareness in skull base surgery. In: Proceedings of the 2024 IEEE International Conference on Robotics and Automation (ICRA); 2024 May 13–17. (2024). 3588–94. doi: 10.1109/ICRA57147.2024.10611187
10
DingHSeenivasanLKilleenBDChoSMUnberathM. Digital twins as a unifying framework for surgical data science: the enabling role of geometric scene understanding. Artif Intell Surg (2024) 4(3):109–38. doi: 10.20517/ais.2024.16
Summary
Keywords
artificial intelligence, digital twin, human–machine cooperation, robotic surgical assistants, robotic surgical procedures, surgery
Citation
Taylor RH (2026) An emerging artificial intelligence-enabled partnership between surgical teams and robotics to enhance patient care. Front Sci 4:1838651. doi: 10.3389/fsci.2026.1838651
Received
25 March 2026
Revised
31 March 2026
Accepted
13 April 2026
Published
07 May 2026
Volume
4 - 2026
Edited and reviewed by
Andrew Gumbs, Hôpital Antoine-Béclère, France
Updates
Copyright
© 2026 Taylor.
This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.
*Correspondence: Russell H. Taylor, rht@jhu.edu
Disclaimer
All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article or claim that may be made by its manufacturer is not guaranteed or endorsed by the publisher.