Why this job stays in the room
Anesthesiologist assistants practice under an anesthesiologist’s direction, and nearly all of the work happens with hands on a patient. You induce and maintain anesthesia, manage the airway, place intravenous and arterial lines, and watch vital signs minute by minute while a surgeon operates. A model can read a monitor trace. It cannot feel a difficult airway, reposition a tube, or decide in four seconds that a blood pressure drop is bleeding rather than a drug effect.
The second reason is accountability. Anesthesia care is delivered inside a care team with a named supervising physician, under state law and hospital credentialing. Software that changes a drug rate is a regulated medical device, not an app. That approval path is slow, and it is one of the clearest brakes on this job listed in the blockers above.
The economics point the same way. The Bureau of Labor Statistics reports median pay of $135,880 for this group and projects employment growth of 21.1% between 2025 and 2035 (BLS, 2025). Demand for surgery is rising faster than anyone is building machines that can stand at the head of the table. If you want the short version of how we weigh all of this, read how we score jobs, or see where the rest of the hospital workforce sits.
What AI does, what it helps with, and what people keep
Share AI can handle on its own: 0%. The tasks closest to that line are paperwork-shaped. Writing the anesthesia record as the case runs, and checking equipment and drug supplies against a standing checklist, are already partly machine work in modern operating rooms. Our Can AI do it? figure is 8 out of 100, measured as the share of task time software can take today; the coverage method explains how that is built.
Share where AI assists a person: 10%. Monitoring depth of anesthesia and responding to changes in a patient’s status is the clearest example. Prediction tools can flag a likely hypotensive episode a few minutes early, and pre-operative chart review can be summarized for you. In both cases a clinician still makes the call and signs for it.
Share that stays with people: 90%. That is intubation and airway rescue, placing invasive lines, administering drugs by hand, and taking part in resuscitation when a case turns. Add the parts nobody automates well: talking a frightened patient through induction, and handing off clearly to the recovery nurse.
What the evidence actually shows
The evidence grade for this job is D. In plain terms, there is no published head-to-head test of an AI system against an anesthesiologist assistant doing this job, so we give no Is it better than a person? number. Nothing has measured a model against a qualified provider across a full case list.
What would settle it is specific, and worth knowing before you believe any headline. A prospective, multi-center trial of closed-loop anesthesia delivery against clinician-managed care, reporting hypotension time, awareness and airway complications. Device clearance for autonomous dosing outside narrow sedation protocols. And a measured result for airway management by a machine, which is the part of the job no demonstration has yet handled in an unselected patient. Until one of those exists, treat claims in either direction as opinion. The quality parity method sets out the bar.
When this could change
Most likely after 2042 (8 in 10 of our scenarios). For what that window is measuring, see the replacement year method.
Two things could pull the date earlier. The first is wider clearance for closed-loop sedation in low-risk, high-volume cases such as endoscopy, which would hand routine maintenance to a device while a clinician covers several rooms. The second is cost: the operating range for software in this job runs far below the cost of a provider, so any system that clears regulators has an obvious buyer in a stretched hospital.
Two things hold it back. Robotics is the big one. Our robotics assessment puts this job in the dexterous humanoid tier, meaning a machine would need human-level hands in a crowded, sterile, unpredictable space before it could do the physical work unaided. The other is liability and supervision rules, which tie anesthesia delivery to a named licensed person. Neither moves on a software release cycle.
How to stay needed
Lean into the parts of this job that sit furthest from a model. Airway management, including the difficult airway, is the skill that keeps you in the room. Invasive line placement under ultrasound is the second. Emergency response during a case, where you read a situation faster than a monitor can classify it, is the third.
Two skills are worth adding. Learn to supervise clinical decision support: know what a hypotension prediction tool is trained on, when it drifts, and how to overrule it without hesitating. And get good at handoff and documentation quality, because that is where automated records go wrong and where a careful provider adds obvious value.
What to do: ask who owns the alarm thresholds on any new monitoring system in your department, and volunteer for that group.
If you are weighing the career, look at the jobs next door: physician assistants, nurse anesthetists and anesthesiologists all share pieces of this work. You can put any two of them side by side on the compare page, scan the wider diagnosing and treating practitioners family, or see how this job sits against the rest of the safest jobs from AI list.