Artificial Intelligence (AI) Agents Versus Agentic AI: What’s the Effect in Spine Surgery?

Article information

Neurospine. 2025;22(2):473-477

Publication date (electronic) : 2025 June 30

doi : https://doi.org/10.14245/ns.2550308.154

¹Department of Orthopaedics, School of Medicine, University of Phayao, Phayao, Thailand

²Division of Research, School of Medicine, University of Phayao, Phayao, Thailand

³Advanced Numerical Optimization Research Group, Department of Mathematics, School of Science, University of Phayao, Thailand

Corresponding Author Wongthawat Liawrungrueang Department of Orthopaedics, School of Medicine, University of Phayao, Phayao, Thailand Email: mint11871@gmail.com

Received 2025 March 9; Revised 2025 April 15; Accepted 2025 April 17.

INTRODUCTION

This editorial aims to clarify the conceptual boundaries between 2 emerging forms of artificial intelligence (AI)—AI agents and agentic AI—and explore how this distinction may influence the future roles of surgeons, surgical planning, and AI in spine care. AI is revolutionizing spine surgery, enhancing diagnostics, surgical planning, and robotic assistance. However, as AI continues to evolve, it is essential to distinguish between AI agents and agentic AI, as they have fundamentally different implications for clinical practice. The term “agentic AI” derives from the broader concept of artificial general intelligence, representing systems capable of autonomy, adaptability, and decision-making beyond predefined rules, distinguishing it from today’s task-specific, or “narrow AI,” applications commonly used in medicine. AI agents are designed for specific tasks such as radiological image analysis, navigation assistance, and robotic augmentation, operating within predefined rules and requiring human oversight. In contrast, agentic AI represents a new paradigm—systems capable of autonomy, real-time decision-making, and self-learning from surgical data. While AI agents are actively transforming surgical workflows, agentic AI remains largely experimental but holds the potential to redefine the role of the surgeon in spine surgery.

AI Agents in Spine Surgery: Current Applications

AI agents have already demonstrated clinically validated benefits across multiple domains of spine surgery. Machine learning models have outperformed traditional radiologists in detecting spinal fractures, degenerative changes, and disc herniations [1-3]. AI-driven segmentation tools assist in spinal alignment analysis and surgical planning, reducing the risk of intraoperative errors [2]. In the realm of surgical navigation, AI-assisted robotic platforms have significantly improved pedicle screw placement accuracy and reduced radiation exposure by minimizing fluoroscopic imaging [2,4,5]. Additionally, AI-powered predictive analytics help estimate fusion success rates, postoperative recovery, and complication risks, allowing for a more personalized approach to spine surgery [2,6]. These AI agents function as assistive technologies, enhancing precision and efficiency while ensuring that the surgeon remains in full control.

Agentic AI: The Next Frontier in Spine Surgery?

Agentic AI, on the other hand, represents a future where AI systems possess greater autonomy in decision-making. Unlike AI agents, agentic AI could potentially adapt in real-time during surgery, modifying the surgical plan based on intraoperative data. For instance, a deep learning system could continuously analyze biomechanical forces and adjust screw placement accordingly, or an AI-powered robotic system could autonomously perform decompression procedures under minimal supervision [2,7,8]. For example, envision a future operating room where an agentic AI-powered robotic system autonomously monitors intraoperative spinal alignment, identifies deviations in real time, and adjusts pedicle screw trajectories during deformity correction—without requiring constant human input. Such capabilities, though currently theoretical, illustrate the potential paradigm shift toward intelligent surgical autonomy. Preliminary studies have explored the feasibility of reinforcement learning AI in surgical simulations, where models improve performance over time by learning from procedural outcomes [2,9]. However, these systems are not yet clinically approved, and significant barriers, including regulatory restrictions, ethical concerns, and liability issues, must be addressed before agentic AI can be safely integrated into spine surgery.

Challenges and Ethical Considerations

The key challenge with agentic AI is ensuring safety, accountability, and transparency. Current U.S. Food and Drug Administration and Conformité Européenne regulations mandate human oversight in AI-assisted surgical systems, preventing AI from making autonomous intraoperative decisions [10]. Deep learning models’ “Black-Box” nature further complicates integration, as it remains challenging to understand how agentic AI reaches certain decisions fully [5,10]. Additionally, medical-legal responsibility becomes an issue: if an autonomous AI system makes an intraoperative decision that leads to complications, who is held accountable, the AI developer, the hospital, or the surgeon? Given these concerns, agentic AI is not yet ready for routine clinical deployment, but ongoing research continues to explore its potential role in spine surgery. We summarize the comparison between AI agents and agentic AI with their clinical impact in Table 1 and provide illustrative figures to depict current and emerging technologies (Figs. 1 and 2).

Table 1.

Summary comparison of artificial intelligence (AI) agents and agentic AI with clinical impact

Fig. 1.

Current artificial intelligence (AI)-assisted technologies in spine surgery. (A) AI-assisted image analysis (e.g., fracture detection using convolutional neural network (CNN), (B) Mazor X robotic surgical platform, (C) Excelsius GPS for spinal navigation, (D) Augmented reality overlays for intraoperative guidance.

Fig. 2.

Agentic artificial intelligence (AI) in other industries. (A) Aviation: autonomous flight control systems, (B) Automotive: self-driving vehicles (e.g., Tesla full self-driving), (C) Manufacturing: adaptive robotic arms with feedback learning, (D) Healthcare: autonomous AI triage systems in emergency care.

The Future: A Hybrid Approach to AI in Spine Surgery

While AI agents and agentic AI represent different stages of technological advancement, their future in spine surgery is likely to be synergistic rather than mutually exclusive [5,11,12]. In the short term, AI agents will continue to refine imaging, navigation, and robotic-assisted surgery, serving as decision-support tools that improve surgical accuracy and efficiency [2,5]. Meanwhile, elements of agentic AI such as real-time adaptability and predictive decision-making may gradually be introduced under human supervision. This hybrid approach will enable intelligent augmentation without compromising patient safety, professional accountability, or clinical ethics. This hybrid approach ensures that AI enhances surgical outcomes without compromising safety, eth-ics, or surgeon expertise.

CONCLUSION

AI is undoubtedly shaping the future of spine surgery, but full autonomy remains a distant goal. The integration of both AI agents and agentic AI represents a continuum of innovation, where evolving technologies can support the surgeon as intelligent collaborators rather than replacements. The most promising path forward is one where AI acts as an intelligent collaborator, augmenting rather than replacing the surgeon. By balancing innovation with patient safety, integrating AI agents, and eventually, agentic AI can lead to safer, more precise, and more personalized spine surgery. As technology advances, spine surgeons must remain at the forefront of AI adoption, ensuring that it aligns with clinical needs while upholding the highest standards of care.

Notes

Conflict of Interest

The author has nothing to disclose.

References

1. Liawrungrueang W, Han I, Cholamjiak W, et al. Artificial intelligence detection of cervical spine fractures using convolutional neural network models. Neurospine 2024;21:833–41.

2. Liawrungrueang W, Cho ST, Sarasombath P, et al. Current trends in artificial intelligence-assisted spine surgery: a systematic review. Asian Spine J 2024;18:146–57.

3. Watanabe K, Aoki Y, Matsumoto M. An application of artificial intelligence to diagnostic imaging of spine disease: estimating spinal alignment from moiré images. Neurospine 2019;16:697–702.

4. Lee S, Jung JY, Mahatthanatrakul A, et al. Artificial intelligence in spinal imaging and patient care: a review of recent advances. Neurospine 2024;21:474–86.

5. Yi S. Artificial intelligence (AI)-robotics started when human capability reached limit, human creativity begin again when the capability of AI-robotics reaches a plateau. Neurospine 2024;21:3–5.

6. Nam KH, Kim DY, Kim DH, et al. Conversational artificial intelligence for spinal pain questionnaire: validation and user satisfaction. Neurospine 2022;19:348–56.

7. Farooq M, Zahra SG. Robotics and artificial intelligence in minimally invasive spine surgery: a bibliometric and visualization analysis. World Neurosurg 2024;190:240–54.

8. Liawrungrueang W, Cholamjiak W, Promsri A, et al. Artificial intelligence for cervical spine fracture detection: a systematic review of diagnostic performance and clinical potential. Global Spine J 2025;15:2547–58.

9. Kalanjiyam GP, Chandramohan T, Raman M, et al. Artificial intelligence: a new cutting-edge tool in spine surgery. Asian Spine J 2024;18:458–71.

10. Joshi G, Jain A, Araveeti SR, et al. FDA-approved artificial intelligence and machine learning (AI/ML)-enabled medical devices: an updated landscape. Electronics 2024;13:498.

11. Zou J, Topol EJ. The rise of agentic AI teammates in medicine. Lancet 2025;405:457.

12. Liawrungrueang W, Cho ST, Cholamjiak W, et al. Performance and clinical implications of machine learning models for detecting cervical ossification of the posterior longitudinal ligament: a systematic review. Asian Spine J 2025;19:148–59.

Article information Continued

This is an open access article distributed under the terms of the Creative Commons Attribution Non-Commercial License (http://creativecommons.org/licenses/by-nc/4.0/) which permits unrestricted non-commercial use, distribution, and reproduction in any medium, provided the original work is properly cited.

Fig. 1.

Fig. 2.

Table 1.

Summary comparison of artificial intelligence (AI) agents and agentic AI with clinical impact

Feature	AI agents	Agentic AI
Definition	AI-driven systems designed for specific, well-defined tasks such as image analysis, surgical navigation, and robotic assistance.	Advanced AI systems with high autonomy, capable of real-time decision-making, adaptive learning, and intraoperative modifications without human intervention.
Autonomy level	Low – Requires direct human supervision, functions within predefined parameters, and executes tasks based on pre-programmed logic.	High – Makes independent decisions, continuously optimizes processes, and modifies surgical execution without predefined constraints.
Primary function	Enhances efficiency, accuracy, and consistency in diagnostics, image processing, preoperative planning, and robotic guidance.	Learns from intraoperative data, adapts surgical plans, and optimizes decisions in real-time based on patient-specific responses and outcomes.
Clinical applications	Widely used in spine surgery, including AI-assisted radiology for fracture detection, robotic-assisted pedicle screw placement, and outcome prediction models.	Experimental – Being explored for applications in autonomous robotic surgery, AI-driven real-time adjustments, and AI-guided surgical execution.
Decision-making capability	Supportive role – Provides recommendations but does not independently alter surgical decisions. The surgeon remains in full control.	Partially or fully autonomous – Can make surgical adjustments based on evolving intraoperative conditions, potentially reducing the surgeon’s direct involvement.
Adaptability and Learning	Static AI models – Operate based on pre-trained datasets and do not update themselves during surgery.	Dynamic AI models – Learn and improve from past procedures, patient responses, and real-time biomechanical feedback.
Regulatory approval	Approved for clinical use in radiology (e.g., AI-assisted spine imaging), robotic-assisted surgery (e.g., Mazor X, ExcelsiusGPS), and navigation systems (e.g., AI-driven augmented reality systems).	Not yet approved due to ethical concerns, lack of interpretability, patient safety risks, and legal liability issues. Requires extensive validation before integration into clinical practice.
Safety and Risk	Low risk – Functions as an assistive tool, operates within surgeon-defined parameters, and reduces human error.	High risk – Concerns include AI-driven surgical errors, liability in case of complications, patient safety, and loss of human oversight in critical situations.
Surgeon’s role	Central – The surgeon remains the primary decision-maker, using AI as a tool to enhance accuracy, efficiency, and consistency.	Evolving – In advanced implementations, the AI may take on an active role in intraoperative decisions, raising concerns about human oversight and accountability.
Limitations	Limited to specific, predefined tasks and lacks the ability to adapt beyond its training data. Struggles with variability in patient-specific anatomy and surgical complexity.	Unpredictability and ethical concerns – Can AI be trusted to make intraoperative decisions without human intervention? How do we ensure safety, reliability, and fairness in AI-driven procedures?
Future potential	Refining AI-assisted navigation and robotic systems for higher accuracy. Enhancing predictive models for patient-specific risk stratification and postoperative recovery. Reducing radiation exposure through AI-optimized intraoperative imaging.	AI-driven real-time intraoperative decision-making and adaptation of surgical plans. Autonomous robotic execution of specific surgical steps. AI-assisted learning from past procedures to continuously refine techniques and optimize outcomes.