AI can help determine which patients reach a spine surgeon, predict surgical risk, generate clinical notes and assist with treatment planning.
But as the technology moves closer to actual care decisions, another question is becoming harder to avoid: When AI influences a decision, who is accountable for it?
For spine surgeons, the answer is still evolving.
The American Medical Association adopted a new policy in June stating that AI should remain an assistive tool rather than an autonomous decision-maker and that physicians should retain meaningful oversight of clinical decisions. At the same time, the AMA is pushing back against efforts that would automatically make physicians responsible for every failure involving an AI tool, particularly when the technology is selected or required by an employer, health system or payer.
The distinction is becoming increasingly relevant in spine, where surgeons are already using or exploring AI across documentation, patient selection, predictive analytics, surgical planning and navigation. Some practices are going further, building AI into the front end of the patient journey before the surgeon ever enters the room.
That expansion is putting physician-level accountability under a new kind of pressure.
The algorithm can advise. The surgeon still has to judge
One principle is emerging clearly across current AI guidance: Clinical judgment is not supposed to disappear because an algorithm enters the workflow.
The FDA’s January 2026 clinical decision-support guidance makes that distinction explicit. Certain clinical decision-support software can fall outside the agency’s medical-device definition when it allows a healthcare professional to independently review the basis for its recommendations rather than rely primarily on the software to make a diagnosis or treatment decision for an individual patient.
The FDA recommends giving clinicians information about the tool’s intended use, required inputs, underlying methodology, validation and relevant limitations.
That matters in spine because AI is moving into decisions with considerably higher stakes than scheduling or administrative automation.
Machine-learning models are already being used to predict spine surgery risks and outcomes. Kevin Huang, MD, a neurosurgeon at Boston-based Massachusetts General Hospital, told Becker’s their value depends heavily on the quality and relevance of the data used to build them.
Some clinically important factors do not fit neatly into structured datasets. Duration of pain, for example, can influence outcomes but may not be consistently captured. Dr. Huang also emphasized the importance of knowing how a model was trained, where it was validated and whether the population resembles the patient sitting in front of the surgeon.
For surgeons, that creates a new layer of responsibility. It is no longer enough to know what an AI model recommends. As these tools move closer to clinical decision-making, physicians may increasingly need to understand why a model produced that recommendation, and when not to follow it.
Accountability can begin before the patient reaches the surgeon
AI’s influence can start even earlier. Spine practices are already exploring algorithmic triage to determine which patients need surgical evaluation and which should first be routed toward therapy, pain management or other care.
Michael Verdon, DO, of Dayton (Ohio) Neurologic Associates, told Becker’s his practice has used an algorithmic triage tool to route patients based on symptom severity. Khalid Odeh, MD, of Michigan Orthopaedic Specialists and Corewell Health in Royal Oak, Mich., has described a model in which neurological deficits would trigger faster surgical evaluation while other patients could be directed toward nonsurgical pathways.
The potential efficiency is significant. So is the accountability question.
If an algorithm is helping decide which patient reaches a surgeon, and how quickly, the critical issue is not merely whether the technology works on average. Practices need to know how neurological red flags are handled, when a human reviews the recommendation, how physicians can override the system and what happens when the algorithm gets it wrong.
That is where physician-level oversight starts to move beyond the operating room.
The clinical note is becoming part of the AI accountability chain
Documentation may appear to be a lower-risk use of AI, but it increasingly connects to everything that follows.
Spine surgeons told Becker’s entering 2026 that ambient AI was already being used to generate office notes. Alex Vaccaro, MD, PhD, president of Philadelphia-based Rothman Orthopaedics, described a broader strategy in which AI-generated structured documentation could eventually feed patient-selection models, personalized risk scores, coding and claim audits.
That means a mistake at the documentation stage may not necessarily stay in the note.
Incorrect or incomplete information could potentially become an input for another tool downstream. Noam Stadlan, MD, of Evanston, Ill.-based Endeavor Health Neurosciences Institute, told Becker’s that although AI can be useful for generating office notes, its output still needs to be checked because of the possibility of inaccurate or fabricated information.
As AI systems become interconnected, the surgeon’s role may increasingly include verifying not only the final recommendation, but also the information being fed into the systems producing it.
Physician accountability does not necessarily mean physician-only accountability
One of the biggest unresolved issues is where a physician’s responsibility should end.
A draft policy under consideration by the Washington Medical Commission illustrates the debate. The proposal would make physicians and other licensees “fully and solely responsible” for clinical judgments and outcomes even when AI tools are involved. It would require independent clinical judgment and critical evaluation of AI outputs. The proposal remains under review; the commission has another interested-parties policy meeting scheduled for Sept. 24.
The AMA has objected to the portion that would place the entire burden on the physician even when an AI tool malfunctioned or its use was embedded into a workflow or mandated by an employer, health system or payer. The organization argues that accountability should instead follow the parties best positioned to understand a technology’s risks and prevent harm through its design, validation and implementation.
That is an important distinction for spine surgeons. Physician-level accountability does not necessarily mean that emerging policy will ultimately make a surgeon personally responsible for every algorithmic failure. That question remains unsettled and can vary by jurisdiction, technology and circumstance.
What is becoming harder to argue, however, is that responsibility disappears simply because software contributed to the decision.
Payers add another layer
Surgeons are also encountering AI they did not choose.
Spine physicians have told Becker’s that insurers are increasingly using algorithmic systems in prior authorization, including tools that review records for required documentation and coverage criteria.
Kasra Ahmadinia, MD, a spine surgeon at Tulsa-based Advanced Orthopedics of Oklahoma, told Becker’s that the dynamic could push physicians toward using their own AI tools to ensure documentation contains the language payer systems expect.
The AMA has separately called for physician oversight when insurers use AI in coverage determinations and opposes autonomous or semiautonomous AI replacing physician review of those decisions.
For spine surgeons, that creates a particularly complicated accountability environment: An individual physician may remain responsible for exercising clinical judgment while algorithms outside that physician’s control increasingly influence whether and when a patient can receive the recommended care.
Knowing which AI you use may no longer be enough
The next stage of AI adoption in spine may therefore require a different level of due diligence.
The AMA’s 2026 framework for evaluating healthcare AI emphasizes five areas: the clinical use case, relevance of training and validation data, risks and mitigation strategies, effectiveness and performance and integration and ongoing monitoring.
How responsibility is divided when AI contributes to an error remains unsettled. But emerging guidance consistently preserves a role for physician review and independent clinical judgment.
At the Becker’s 32nd Annual Meeting: The Business and Operations of ASCs, taking place October 29-31 in Chicago, ASC leaders, surgeons and healthcare executives will explore strategies to drive growth, enhance operational performance, navigate reimbursement challenges and prepare for the future of ambulatory surgery. Apply for complimentary registration now.
