Will AI Replace Anesthesiologists?
No. This question has already been tested in the real market, not just in theory, and the answer came back clearly: a company tried to automate routine sedation with a machine over a decade ago, and it failed commercially even after getting regulatory approval. AI monitoring tools today are better than that machine ever was, and they are still built as decision support, not replacement.
Anesthesiology holds up well against automation because the job is defined by moment-to-moment judgment under changing conditions, exactly the kind of live, high-accountability decision-making that current systems are not certified to make unsupervised.
What the data actually says
BLS projects 3% employment growth for anesthesiologists from 2024 to 2034, adding roughly 1,400 new positions, tracked as part of the broader "Physicians and Surgeons" category which also includes about 23,600 openings a year across all physician and surgeon roles (BLS). For comparison, nurse anesthetists (tracked under a separate BLS category alongside nurse midwives and nurse practitioners) are projected to grow 35% from 2024 to 2034, much faster than average, with about 32,700 openings a year across that combined group (BLS).
Neither number shows a field being displaced. Both show steady or fast growth, consistent with an aging population needing more surgical and procedural care, which is the same driver behind most of healthcare's workforce numbers.
The Sedasys history, and why it matters here
The Sedasys history matters because it's the closest thing to a real-world experiment in automating this exact job, and it happened years before generative AI existed.
Johnson & Johnson developed Sedasys, a computerized system designed to administer propofol sedation for routine procedures like colonoscopies without a dedicated anesthesiologist present. The system won FDA approval in 2013, after an earlier rejection in 2010, and was explicitly built to reduce or remove the need for anesthesia specialists during specific low-risk procedures (Anesthesia Experts).
It faced serious, organized professional opposition before it even reached the market. The American Society of Anesthesiologists and the American College of Anesthesiologists lobbied against Sedasys during its development on safety grounds, arguing sedation was too delicate to hand to an automated system (anesthesiology industry coverage). J&J eventually modified the product's ambitions and both professional groups came around to supporting a narrower version of it.
None of that saved it commercially. J&J pulled Sedasys from the market only three years after approval, citing unexpectedly slow sales and company-wide cost cutting, not a safety failure (Anesthesiology News). Adoption was described as modest at best, with only a handful of providers actually buying and using the system before it was discontinued (MassDevice).
The lesson from Sedasys isn't that automation in anesthesia is technically impossible. It's that the market, the profession, and the liability structure around anesthesia rejected a narrow automation attempt even for the lowest-risk procedures it targeted, well before AI made monitoring tools meaningfully better than they were in 2013.
Which tasks are exposed
Exposed: pre-op documentation, routine dosing calculations that follow an established protocol for a standard, low-complexity patient, and post-op note generation.
Newer research supports AI as a monitoring aid here specifically. A 2025 study built a hybrid model combining several machine-learning architectures to predict anesthesia depth from drug-infusion history, improving predictive accuracy over older methods when tested against a public anesthesia database (PubMed). Separate 2025 research applying artificial neural networks to EEG-based depth-of-anesthesia classification found the models outperformed traditional monitors on that specific classification task (anesthesia AI review).
Which tasks are protected, and why
Protected: the moment-to-moment judgment during a procedure when a patient's vitals shift unexpectedly, dosing decisions for any patient with complicating factors (age, weight extremes, drug interactions, comorbidities), and the accountability that comes with being the specific licensed person responsible if something goes wrong mid-procedure.
The AI depth-of-anesthesia research above is explicit about its own scope: the goal is to give anesthesiologists better information to base a decision on, not to make the decision itself (hybrid AI anesthesia depth study). Every current model in this space is framed as decision support layered under a human, never as an independent actor administering drugs to a patient without supervision.
What is already happening
The direction of real research since Sedasys has moved toward augmenting the anesthesiologist's information, not replacing their function. A broader 2025 literature review of machine learning in anesthesia covers automated drug titration and real-time physiologic optimization as active research areas, but frames all of it as tools that still sit inside a supervised clinical workflow (PMC systematic review).
Nothing in current regulatory practice allows an AI system to independently administer anesthesia to a patient without a licensed anesthesiologist or certified registered nurse anesthetist present and responsible. That is the same structural protection that shields surgery: liability sits with a specific licensed person, and no system has been built, approved, or trusted to take that liability on itself.
What to do about it
- Get familiar with AI-assisted depth-of-anesthesia monitoring now. These tools are showing up in research and early clinical use as genuinely useful decision support, and being an early, skilled adopter is a stronger position than resisting the tooling.
- Don't worry about Sedasys-style automation repeating. The commercial and professional failure of that specific product is a real historical data point against narrow automation succeeding here, even for low-risk procedures.
- Understand where nurse anesthetists fit into your practice's growth. BLS shows much faster growth for nurse anesthetists than for anesthesiologists, which reflects a broader anesthesia-care team model, not automation.
- Keep documentation tools out of the way of your actual monitoring attention. The exposed tasks here (notes, routine calculations) are worth automating specifically so more attention goes to the protected task: watching the patient.
Keep the skills that keep you employed
The tasks you can still do without leaning on AI are what make you hard to replace here. The free 5-Day AI Reset is a five-email course built around exactly that: Day 2 has you take one task back and do it unassisted. One small change per day, and it stays useful no matter which way anesthesiologists moves.
The tasks you keep decide how replaceable you are
The tasks you can still do without leaning on AI are what make you hard to replace here. The free 5-Day AI Reset is a five-email course built around exactly that: Day 2 has you take one task back and do it unassisted. One small change per day, and it stays useful no matter which way the profession evolves.
Frequently asked questions
Will AI replace anesthesiologists?
No. Anesthesiology requires continuous, real-time judgment under changing patient conditions, and no current AI system is certified to administer anesthesia independently. BLS projects steady 3% employment growth for anesthesiologists through 2034 (BLS).
What was Sedasys and why did it fail?
Sedasys was a Johnson & Johnson system approved by the FDA in 2013 to automate sedation for routine procedures like colonoscopies without a dedicated anesthesiologist. It was withdrawn from the market in 2016 due to weak sales and company cost-cutting, not a safety failure, after facing strong professional opposition and limited hospital adoption (Anesthesiology News).
Can AI monitor anesthesia depth better than a human?
Recent research shows AI models, including neural networks analyzing EEG data, can outperform traditional monitors on specific depth-of-anesthesia classification tasks (2025 study). These tools are built to inform an anesthesiologist's decision, not to act on their own.
Is anesthesiology a safe career choice given AI?
Yes. The field combines licensed accountability, physical presence, and continuous real-time judgment, the same combination that protects surgery and other hands-on clinical roles from automation. The one real-world attempt to automate part of this job commercially failed.
Do nurse anesthetists face more or less AI risk than anesthesiologists?
Neither group shows displacement risk in current data. BLS actually projects much faster growth for nurse anesthetists (35% through 2034) than for anesthesiologists (3%), reflecting expanded anesthesia-care team models rather than automation replacing either role (BLS).
Want your own number instead of an industry average? Take the free How AI-Proof Is Your Job? assessment and see how your actual day-to-day tasks score.