Will AI replace Physician Assistants?

Not as a job, but it is already doing parts of the work. Across the 28 official task statements scored for Physician Assistants (United States, SOC 29-1071), 2% of the importance-weighted core work is made of tasks today's AI could already do most of. The overall exposure score is 15 out of 100 (range 12–20, band: minimal). That is a statement about tasks, not about headcount: this measures what AI could do, not whether any employer adopts it, whether the law allows it, or whether doing the routine parts faster creates more demand for the human parts. Figures are from release 2026-q4.1.

Which tasks in “Physician Assistants” can AI already do?

The highest-scoring tasks in release 2026-q4.1 are: “Order medical and laboratory supplies and equipment” (75/100, high); “Interpret diagnostic test results for deviations from normal” (47/100, partial); “Obtain, compile, and record patient medical data, including health history, progress notes, and results of physical examination” (42/100, partial). Each score comes from five published 0–4 ratings turned into a number by a published formula, and each carries the model's one-sentence reason on the page.

Which tasks in “Physician Assistants” stay human?

About 85% of this job's task weight sits in work that scores low for AI exposure. The lowest-scoring tasks in release 2026-q4.1 are: “Administer anesthetic, adjuvant, or accessory drugs under the direction of an anesthesiologist” (0/100, minimal); “Control anesthesia levels during procedures” (0/100, minimal); “Administer blood, blood products, or supportive fluids” (0/100, minimal). Low scores usually mean the task needs a body in a room, a legally accountable human, or trust built in real time. Those are the three things the scoring rubric treats as gates rather than obstacles.

What should someone working in “Physician Assistants” do about AI?

Start from the ledger rather than the headline: 2% of this job's weighted core work is exposed, and roughly 85% is not. The practical move is to spend more of your week on the tasks that score low, the ones above, and to get fluent at directing AI through the tasks that score high, because those are the parts that change whether or not you are ready for them. This page does not predict your job, and nothing here is career advice tailored to you: the score describes the occupation, not the person.

How is the AI exposure score for Physician Assistants calculated?

Each official task statement for the occupation is rated on five published 0–4 dimensions (output replicability, physical embodiment, licensed accountability, real-time human trust, and data availability) by claude-opus-5 using scoring prompt task_scoring_v1.0. The model never writes the score; a published formula turns the five ratings into a 0–100 number, so every score can be recomputed by hand. The occupation figure is the importance-weighted mean across 28 scored tasks. The prompt, the rubric, the formula and the full dataset are published at https://futureproof.collab365.com/method and https://futureproof.collab365.com/data/2026-q4.1 under CC BY 4.0.

US dataswitch to UK

Physician Assistants

making tentative diagnoses and decisions about management and treatment of patients, providing airway management interventions, tracheal intubation, fiber optics or ventilary support and responding to emergency situations by providing cardiopulmonary resuscitation. If that's your week, this page is about your job.

The honest answer

AI changes the edges of this job, not the middle: examining patients to obtain information about their physical condition is work software can't reach.

Your move: what you can actually do about this ↓

Where the work sits, by task weight

shifting to AI: 2%
changing shape: 13%
staying human: 85%

These bars are tasks changing hands, not people being counted out. The ledger below shows which.

Whole-job exposure score 15 out of 100 (12–20 allowing for uncertainty): minimal exposure, across 28 scored tasks. The number is the support for the sentence above it, not a headline about anyone’s future.

How we know this

What is measured: Every published task statement for physician assistants is rated on five dimensions: can a model produce the output, does the work need a body in a room, does it need a legally accountable person, does it depend on a person being trusted in the moment, and how much data exists. A published formula turns those five ratings into the score; the model never writes the number.

How the bar is built: Each task’s share of the bar is its published importance weight, so a task you do all day counts for more than one you do twice a year.

Release: 2026-q4.1, scores computed 2026-08-04. Read the full method.

What shifts is ordering medical and laboratory supplies and equipment: the overhead at the edges, not the middle you trained for.

Your week, as this page understands it

Provide healthcare services typically performed by a physician, under the supervision of a physician. Conduct complete physicals, provide treatment, and counsel patients. May, in some cases, prescribe medication. Must graduate from an accredited educational program for physician assistants. The job title says “physician assistants”. The real job is the part underneath: examining patients to obtain information about their physical condition. That is the thing someone has to be right about.

The exposed part of this job is specific, and we won’t pretend it is coming back. But physician assistants is not one task. It is 28 scored ones, and they are nowhere near equally exposed. The most durable of them, on this evidence, is examining patients to obtain information about their physical condition, and the ledger below shows exactly why.

Where the work sits, by task weight

shifting to AI: 2%
changing shape: 13%
staying human: 85%

These bars are tasks changing hands, not people being counted out. The ledger below shows which.

How we know this

How the bar is built: Each task’s share of the bar is its published importance weight, so a task you do all day counts for more than one you do twice a year.

Release: 2026-q4.1, scores computed 2026-08-04. Read the full method.

Your job, task by task

These are the official task statements for this occupation, in plain English, sorted by what the evidence says is happening to each one. The official wording sits under every line so you can check the rewrite against it.

1 of this occupation's scored task statements carry a score that was measured under a different occupation's context, because the statement is shared between jobs and has only been scored once so far. Each one names the occupation it was measured under in the free routes below; none is presented as a measurement for this job.
Task clusters are not derived in this release, so the task-cluster field is empty and no Collab365 Space routing is attached to this occupation yet.

Shifting to AI

1 task

Tasks today’s tools can already do most of. This is the part we will not soften: where these rows are the bulk of your week, the week changes.

Ordering medical and laboratory supplies and equipment
This is reading one thing and writing another: medical and laboratory supplies and equipment in, a record out. That is the shape today's tools are built for.
importance 3 · Core
Source: “Order medical and laboratory supplies and equipment.” (O*NET task statement)
How this row was scored
Exposure score: 75 out of 100 (71–79 allowing for uncertainty): high exposure, high confidence.
Why it sits in this group: reading one thing and writing another; the same decision, made over and over; mistakes that are cheap to catch.
The rating behind it: Reordering supplies from stock records is routine ordering work that software handles end to end.
The five ratings: output a model can produce 3/4 · needs a body in a room 0/4 · needs an accountable person 0/4 · needs to be trusted in the moment 0/4 · how much data exists 3/4.

Changing shape

3 tasks

Tasks where the machine takes the producing and a person keeps the checking, the deciding, or the answering-for-it. For most jobs this is the biggest group, and it is where "transformation, not termination" is literally visible.

Interpreting diagnostic test results for deviations from normal
The software now makes the first pass at diagnostic test results, but someone has to be answerable for the result, and it cannot be the software. So the job becomes checking and deciding rather than producing.
importance 5 · Core
Source: “Interpret diagnostic test results for deviations from normal.” (O*NET task statement)
How this row was scored
Exposure score: 47 out of 100 (40–54 allowing for uncertainty): partial exposure, medium confidence.
Why it sits in this group: reading one thing and writing another; the same decision, made over and over; someone qualified has to answer for it.
The rating behind it: Spotting results outside the normal range suits software, though a licensed clinician must judge what they mean.
The five ratings: output a model can produce 3/4 · needs a body in a room 0/4 · needs an accountable person 3/4 · needs to be trusted in the moment 0/4 · how much data exists 3/4.
Obtaining, compiling and recording patient medical data, including health history, progress notes and results of physical examination
The software now makes the first pass at patient medical data, including health history, but the deciding part still needs a person. So the job becomes checking and deciding rather than producing.
importance 5 · Core
Source: “Obtain, compile, and record patient medical data, including health history, progress notes, and results of physical examination.” (O*NET task statement)
How this row was scored
Exposure score: 42 out of 100 (38–46 allowing for uncertainty): partial exposure, high confidence.
Why it sits in this group: reading one thing and writing another; the same decision, made over and over; mistakes that are cheap to catch.
The rating behind it: Note-taking tools already turn a consultation into a written record, and the writing itself carries no licensing restriction.
The five ratings: output a model can produce 3/4 · needs a body in a room 1/4 · needs an accountable person 1/4 · needs to be trusted in the moment 1/4 · how much data exists 3/4.
Participating in seminars, workshops or other professional activities to keep abreast of developments in anesthesiology
The software now makes the first pass at seminars, workshops or other professional activities, but the deciding part still needs a person. So the job becomes checking and deciding rather than producing.
importance 4 · Core
Source: “Participate in seminars, workshops, or other professional activities to keep abreast of developments in anesthesiology.” (O*NET task statement)
How this row was scored
Exposure score: 41 out of 100 (34–48 allowing for uncertainty): partial exposure, medium confidence.
Why it sits in this group: the same decision, made over and over; mistakes that are cheap to catch.
The rating behind it: Software can summarize new developments in the field, but attending professional events is still a person turning up.
The five ratings: output a model can produce 2/4 · needs a body in a room 1/4 · needs an accountable person 0/4 · needs to be trusted in the moment 1/4 · how much data exists 4/4.

Staying human

24 tasks

Tasks that stay with a person, because they happen in the physical world, because the rules need someone accountable, or because the value is that a specific person does them.

Making tentative diagnoses and decisions about management and treatment of patients
The rules require a named, qualified person to answer for tentative diagnoses, and that person cannot be a piece of software.
importance 5 · Core
Source: “Make tentative diagnoses and decisions about management and treatment of patients.” (O*NET task statement)
How this row was scored
Exposure score: 30 out of 100 (23–37 allowing for uncertainty): low exposure, medium confidence.
Why it sits in this group: reading one thing and writing another; the same decision, made over and over; someone qualified has to answer for it.
The rating behind it: A first-pass diagnosis from recorded findings is something AI does well, but law requires a licensed clinician to decide treatment.
The five ratings: output a model can produce 3/4 · needs a body in a room 1/4 · needs an accountable person 3/4 · needs to be trusted in the moment 1/4 · how much data exists 3/4.
Prescribing therapy or medication with physician approval
The rules require a named, qualified person to answer for therapy, and that person cannot be a piece of software.
importance 5 · Core
Source: “Prescribe therapy or medication with physician approval.” (O*NET task statement)
How this row was scored
Exposure score: 24 out of 100 (20–28 allowing for uncertainty): low exposure, high confidence.
Why it sits in this group: the same decision, made over and over; someone qualified has to answer for it.
The rating behind it: Prescribing is legally reserved to licensed prescribers, even though treatment guidelines are well documented and easy to look up.
The five ratings: output a model can produce 2/4 · needs a body in a room 0/4 · needs an accountable person 4/4 · needs to be trusted in the moment 1/4 · how much data exists 3/4.
Examining patients to obtain information about their physical condition
This work happens in the physical world: patients, in a real place. Software cannot follow it there.
importance 5 · Core
Source: “Examine patients to obtain information about their physical condition.” (O*NET task statement)
How this row was scored
Exposure score: 0 out of 100 (0–4 allowing for uncertainty): minimal exposure, high confidence.
Why it sits in this group: work that happens in the physical world; someone qualified has to answer for it.
The rating behind it: Examining a patient means physically checking their body, which software cannot do.
The five ratings: output a model can produce 0/4 · needs a body in a room 4/4 · needs an accountable person 3/4 · needs to be trusted in the moment 1/4 · how much data exists 2/4.
Administering or ordering diagnostic tests
This work happens in the physical world: diagnostic tests, in a real place. Software cannot follow it there.
importance 5 · Core
Source: “Administer or order diagnostic tests, such as x-ray, electrocardiogram, and laboratory tests.” (O*NET task statement)
How this row was scored
Exposure score: 18 out of 100 (11–25 allowing for uncertainty): minimal exposure, medium confidence.
Why it sits in this group: the same decision, made over and over; work that happens in the physical world; someone qualified has to answer for it.
The rating behind it: Choosing which tests to run is documented knowledge, but carrying them out is hands-on and reserved to clinicians.
The five ratings: output a model can produce 2/4 · needs a body in a room 2/4 · needs an accountable person 3/4 · needs to be trusted in the moment 0/4 · how much data exists 3/4.
Instructing and counseling patients about prescribed therapeutic regimens
The rules require a named, qualified person to answer for patients, and that person cannot be a piece of software.
importance 5 · Core
Source: “Instruct and counsel patients about prescribed therapeutic regimens, normal growth and development, family planning, emotional problems of daily living, and health maintenance.” (O*NET task statement)
How this row was scored
Exposure score: 18 out of 100 (11–25 allowing for uncertainty): minimal exposure, medium confidence.
Why it sits in this group: the same decision, made over and over; someone qualified has to answer for it; the value is that a specific person does it.
The rating behind it: Health advice can be drafted automatically, but patients follow guidance they trust from a person who knows them.
The five ratings: output a model can produce 2/4 · needs a body in a room 1/4 · needs an accountable person 2/4 · needs to be trusted in the moment 3/4 · how much data exists 3/4.
Verifying availability of operating room supplies
This work happens in the physical world: availability of operating room supplies, in a real place. Software cannot follow it there.
importance 5 · Core
Source: “Verify availability of operating room supplies, medications, and gases.” (O*NET task statement)
How this row was scored
Exposure score: 33 out of 100 (26–40 allowing for uncertainty): low exposure, medium confidence.
Why it sits in this group: the same decision, made over and over; mistakes that are cheap to catch; work that happens in the physical world.
The rating behind it: Stock levels can be checked from a system, though someone still has to walk the room and confirm what is there.
The five ratings: output a model can produce 3/4 · needs a body in a room 2/4 · needs an accountable person 1/4 · needs to be trusted in the moment 0/4 · how much data exists 3/4.

Show the other 18 tasks

Collecting and documenting patients' pre-anesthetic health histories
staying human
The rules require a named, qualified person to answer for patients' pre-anesthetic health histories, and that person cannot be a piece of software.
importance 5 · Supplemental
Source: “Collect and document patients' pre-anesthetic health histories.” (O*NET task statement)
How this row was scored
Exposure score: 36 out of 100 (29–43 allowing for uncertainty): low exposure, medium confidence.
Why it sits in this group: reading one thing and writing another; the same decision, made over and over; someone qualified has to answer for it.
The rating behind it: Structured health questionnaires and automatic write-ups already collect and record histories well, with a clinician confirming them.
The five ratings: output a model can produce 3/4 · needs a body in a room 1/4 · needs an accountable person 2/4 · needs to be trusted in the moment 1/4 · how much data exists 3/4.
Providing clinical instruction, supervision or training to staff in areas, anesthesia practices
staying human
This work happens in the physical world: clinical instruction, supervision, in a real place. Software cannot follow it there.
importance 4 · Supplemental
Source: “Provide clinical instruction, supervision or training to staff in areas such as anesthesia practices.” (O*NET task statement)
How this row was scored
Exposure score: 15 out of 100 (8–22 allowing for uncertainty): minimal exposure, medium confidence.
Why it sits in this group: the same decision, made over and over; work that happens in the physical world; someone qualified has to answer for it; the value is that a specific person does it.
The rating behind it: Teaching materials can be drafted automatically, but clinical training is largely shown in person at the bedside.
The five ratings: output a model can produce 2/4 · needs a body in a room 2/4 · needs an accountable person 2/4 · needs to be trusted in the moment 2/4 · how much data exists 3/4.
Supervising and coordinating activities of technicians and technical assistants
staying human
This work happens in the physical world: activities of technicians, in a real place. Software cannot follow it there.
importance 4 · Core
Source: “Supervise and coordinate activities of technicians and technical assistants.” (O*NET task statement)
How this row was scored
Exposure score: 9 out of 100 (5–13 allowing for uncertainty): minimal exposure, high confidence.
Why it sits in this group: work that happens in the physical world; the value is that a specific person does it.
The rating behind it: Supervising technicians is a person-to-person job built on knowing their work, though task lists can be drafted automatically.
The five ratings: output a model can produce 1/4 · needs a body in a room 2/4 · needs an accountable person 0/4 · needs to be trusted in the moment 3/4 · how much data exists 2/4.
Monitoring and documenting patients' progress during post-anesthesia period
staying human
This work happens in the physical world: patients' progress during post-anesthesia period, in a real place. Software cannot follow it there.
importance 4 · Supplemental
Source: “Monitor and document patients' progress during post-anesthesia period.” (O*NET task statement)
How this row was scored
Exposure score: 8 out of 100 (1–15 allowing for uncertainty): minimal exposure, medium confidence.
Why it sits in this group: the same decision, made over and over; work that happens in the physical world; someone qualified has to answer for it.
The rating behind it: Recovery-room notes are written work, but watching a patient wake safely means being in the room.
The five ratings: output a model can produce 2/4 · needs a body in a room 3/4 · needs an accountable person 3/4 · needs to be trusted in the moment 1/4 · how much data exists 3/4.
Visiting and observing patients on hospital rounds or house calls
staying human
This work happens in the physical world: patients, in a real place. Software cannot follow it there.
importance 5 · Core
Source: “Visit and observe patients on hospital rounds or house calls, updating charts, ordering therapy, and reporting back to physician.” (O*NET task statement)
How this row was scored
Exposure score: 6 out of 100 (2–10 allowing for uncertainty): minimal exposure, high confidence.
Why it sits in this group: the same decision, made over and over; work that happens in the physical world; someone qualified has to answer for it; the value is that a specific person does it.
The rating behind it: Ward rounds and house calls mean being at the bedside, even though the chart updates afterwards are written work.
The five ratings: output a model can produce 2/4 · needs a body in a room 3/4 · needs an accountable person 3/4 · needs to be trusted in the moment 2/4 · how much data exists 3/4.
Performing therapeutic procedures, such as injections, immunizations, suturing and wound care and infection management
staying human
This work happens in the physical world: therapeutic procedures, in a real place. Software cannot follow it there.
importance 5 · Core
Source: “Perform therapeutic procedures, such as injections, immunizations, suturing and wound care, and infection management.” (O*NET task statement)
How this row was scored
Exposure score: 0 out of 100 (0–4 allowing for uncertainty): minimal exposure, high confidence.
Why it sits in this group: work that happens in the physical world; someone qualified has to answer for it.
The rating behind it: Injections, stitches and wound care all need trained hands on the patient.
The five ratings: output a model can produce 0/4 · needs a body in a room 4/4 · needs an accountable person 3/4 · needs to be trusted in the moment 1/4 · how much data exists 2/4.
Providing physicians with assistance during surgery or complicated medical procedures
staying human
This work happens in the physical world: physicians, in a real place. Software cannot follow it there.
importance 4 · Core
Source: “Provide physicians with assistance during surgery or complicated medical procedures.” (O*NET task statement)
How this row was scored
Exposure score: 0 out of 100 (0–4 allowing for uncertainty): minimal exposure, high confidence.
Why it sits in this group: work that happens in the physical world; someone qualified has to answer for it.
The rating behind it: Helping during an operation is physical work in the operating room.
The five ratings: output a model can produce 0/4 · needs a body in a room 4/4 · needs an accountable person 3/4 · needs to be trusted in the moment 1/4 · how much data exists 1/4.
Collecting samples or specimens for diagnostic testing
staying human
This work happens in the physical world: samples, in a real place. Software cannot follow it there.
importance 4 · Supplemental
Source: “Collect samples or specimens for diagnostic testing.” (O*NET task statement)
How this row was scored
Exposure score: 0 out of 100 (0–4 allowing for uncertainty): minimal exposure, high confidence.
Why it sits in this group: the same decision, made over and over; work that happens in the physical world.
The rating behind it: Taking blood and other samples from a patient is physical work.
The five ratings: output a model can produce 0/4 · needs a body in a room 4/4 · needs an accountable person 1/4 · needs to be trusted in the moment 0/4 · how much data exists 3/4.
Providing airway management interventions, tracheal intubation, fiber optics or ventilary support
staying human
This work happens in the physical world: airway management interventions, tracheal intubation, in a real place. Software cannot follow it there.
importance 5 · Core
Source: “Provide airway management interventions including tracheal intubation, fiber optics, or ventilary support.” (O*NET task statement)
How this row was scored
Exposure score: 0 out of 100 (0–4 allowing for uncertainty): minimal exposure, high confidence.
Why it sits in this group: work that happens in the physical world; someone qualified has to answer for it.
The rating behind it: Managing a patient airway is a physical procedure performed at the bedside.
The five ratings: output a model can produce 0/4 · needs a body in a room 4/4 · needs an accountable person 3/4 · needs to be trusted in the moment 0/4 · how much data exists 1/4.
Responding to emergency situations by providing cardiopulmonary resuscitation
staying human
This work happens in the physical world: emergency situations, in a real place. Software cannot follow it there.
importance 5 · Core
Source: “Respond to emergency situations by providing cardiopulmonary resuscitation (CPR), basic cardiac life support (BLS), advanced cardiac life support (ACLS), or pediatric advanced life support (PALS).” (O*NET task statement)
How this row was scored
Exposure score: 0 out of 100 (0–4 allowing for uncertainty): minimal exposure, high confidence.
Why it sits in this group: work that happens in the physical world; someone qualified has to answer for it.
The rating behind it: Resuscitation is hands-on emergency care that has to happen at the patient.
The five ratings: output a model can produce 0/4 · needs a body in a room 4/4 · needs an accountable person 3/4 · needs to be trusted in the moment 1/4 · how much data exists 2/4.
Pretesting and calibrating anesthesia delivery systems and monitors
staying human
This work happens in the physical world: anesthesia delivery systems, in a real place. Software cannot follow it there.
importance 4 · Core
Source: “Pretest and calibrate anesthesia delivery systems and monitors.” (O*NET task statement)
How this row was scored
Exposure score: 0 out of 100 (0–4 allowing for uncertainty): minimal exposure, high confidence.
Why it sits in this group: work that happens in the physical world; someone qualified has to answer for it.
The rating behind it: Testing and calibrating anesthesia machines means handling the equipment itself.
The five ratings: output a model can produce 0/4 · needs a body in a room 4/4 · needs an accountable person 2/4 · needs to be trusted in the moment 0/4 · how much data exists 2/4.
Assisting anesthesiologists in monitoring of patients
staying human
This work happens in the physical world: anesthesiologists, in a real place. Software cannot follow it there.
importance 5 · Supplemental
Source: “Assist anesthesiologists in monitoring of patients, including electrocardiogram (EKG), direct arterial pressure, central venous pressure, arterial blood gas, hematocrit, or routine measurement of temperature, respiration, blood pressure or heart rate.” (O*NET task statement)
How this row was scored
Exposure score: 0 out of 100 (0–4 allowing for uncertainty): minimal exposure, high confidence.
Why it sits in this group: the same decision, made over and over; work that happens in the physical world; someone qualified has to answer for it.
The rating behind it: Taking readings and watching a patient under anesthesia means standing at the table with them.
The five ratings: output a model can produce 1/4 · needs a body in a room 4/4 · needs an accountable person 3/4 · needs to be trusted in the moment 0/4 · how much data exists 3/4.
Assisting in the provision of advanced life support techniques
staying human
This work happens in the physical world: the provision of advanced life support techniques, in a real place. Software cannot follow it there.
importance 5 · Supplemental
Source: “Assist in the provision of advanced life support techniques including those procedures using high frequency ventilation or intra-arterial cardiovascular assistance devices.” (O*NET task statement)
How this row was scored
Exposure score: 0 out of 100 (0–4 allowing for uncertainty): minimal exposure, high confidence.
Why it sits in this group: work that happens in the physical world; someone qualified has to answer for it.
The rating behind it: Advanced life support equipment has to be applied and managed by hand.
The five ratings: output a model can produce 0/4 · needs a body in a room 4/4 · needs an accountable person 3/4 · needs to be trusted in the moment 0/4 · how much data exists 2/4.
Assisting anesthesiologists in performing anesthetic procedures
staying human
This work happens in the physical world: anesthesiologists, in a real place. Software cannot follow it there.
importance 4 · Supplemental
Source: “Assist anesthesiologists in performing anesthetic procedures, such as epidural or spinal injections.” (O*NET task statement)
How this row was scored
Exposure score: 0 out of 100 (0–4 allowing for uncertainty): minimal exposure, high confidence.
Why it sits in this group: work that happens in the physical world; someone qualified has to answer for it.
The rating behind it: Helping place an epidural or spinal injection is physical work at the patient.
The five ratings: output a model can produce 0/4 · needs a body in a room 4/4 · needs an accountable person 3/4 · needs to be trusted in the moment 0/4 · how much data exists 2/4.
Assisting in the application of monitoring techniques
staying human
This work happens in the physical world: the application of monitoring techniques, in a real place. Software cannot follow it there.
importance 4 · Supplemental
Source: “Assist in the application of monitoring techniques, such as pulmonary artery catheterization, electroencephalographic spectral analysis, echocardiography, or evoked potentials.” (O*NET task statement)
How this row was scored
Exposure score: 0 out of 100 (0–4 allowing for uncertainty): minimal exposure, high confidence.
Why it sits in this group: work that happens in the physical world; someone qualified has to answer for it.
The rating behind it: Fitting catheters and monitoring probes to a patient requires trained hands.
The five ratings: output a model can produce 0/4 · needs a body in a room 4/4 · needs an accountable person 3/4 · needs to be trusted in the moment 0/4 · how much data exists 2/4.
Administering blood, blood products or supportive fluids
staying human
This work happens in the physical world: blood, blood products or supportive fluids, in a real place. Software cannot follow it there.
importance 5 · Supplemental
Source: “Administer blood, blood products, or supportive fluids.” (O*NET task statement)
How this row was scored
Exposure score: 0 out of 100 (0–4 allowing for uncertainty): minimal exposure, high confidence.
Why it sits in this group: work that happens in the physical world; someone qualified has to answer for it.
The rating behind it: Giving blood or fluids is a physical procedure carried out at the patient.
The five ratings: output a model can produce 0/4 · needs a body in a room 4/4 · needs an accountable person 3/4 · needs to be trusted in the moment 0/4 · how much data exists 2/4.
Controlling anesthesia levels during procedures
staying human
This work happens in the physical world: anesthesia levels during procedures, in a real place. Software cannot follow it there.
importance 5 · Supplemental
Source: “Control anesthesia levels during procedures.” (O*NET task statement)
How this row was scored
Exposure score: 0 out of 100 (0–4 allowing for uncertainty): minimal exposure, high confidence.
Why it sits in this group: work that happens in the physical world; someone qualified has to answer for it.
The rating behind it: Adjusting anesthesia during an operation is a hands-on act reserved by law to qualified staff.
The five ratings: output a model can produce 0/4 · needs a body in a room 4/4 · needs an accountable person 4/4 · needs to be trusted in the moment 0/4 · how much data exists 2/4.
Administering anesthetic, adjuvant or accessory drugs under the direction of an anesthesiologist
staying human
This work happens in the physical world: anesthetic, adjuvant or accessory drugs under the direction, in a real place. Software cannot follow it there.
importance 4 · Supplemental
Source: “Administer anesthetic, adjuvant, or accessory drugs under the direction of an anesthesiologist.” (O*NET task statement)
How this row was scored
Exposure score: 0 out of 100 (0–4 allowing for uncertainty): minimal exposure, high confidence.
Why it sits in this group: the same decision, made over and over; work that happens in the physical world; someone qualified has to answer for it.
The rating behind it: Giving drugs to a patient is a hands-on act the law reserves to qualified staff.
The five ratings: output a model can produce 0/4 · needs a body in a room 4/4 · needs an accountable person 4/4 · needs to be trusted in the moment 0/4 · how much data exists 3/4.

What this job pays, and how many people do it

Median pay: $135,880a year, the middle of the range, so half earn more and half earn less.bls-oews, 2025 · May 2025 estimates (national_M2025_dl.xlsx)
How we know this
Source: bls-oews
Reference period: May 2025 estimates (national_M2025_dl.xlsx)
Rounding: Shown as published.
People doing this job: 162,150in the US, 2025.bls-oews · May 2025 estimates (national_M2025_dl.xlsx)

What is deliberately not here: a forecast of how many of these jobs exist in ten years. Where an official projection exists for a market we publish it with its vintage; where it does not, we leave the space empty rather than borrow the other country’s number.

Why this is shifting

The reason is boringly specific. Most of what is shifting here is reading one thing and writing another: medical and laboratory supplies and equipment in, a record out. The rows above are exactly that shape: ordering medical and laboratory supplies and equipment and interpreting diagnostic test results for deviations from normal. What it cannot do is be answerable: patients need a named person the rules will accept, and software cannot be that person. Which is why this page talks about your tasks changing, not your job ending.

Your move

Over a pint: what I’d tell you if you were my friend

Start with what does not change: examining patients to obtain information about their physical condition is the middle of this job, and the evidence on this page says it stays with a person.

So, given all that: 2% of this job's task weight sits in rows the software is already learning, 13% in rows that change shape rather than disappear, and 85% in rows it is nowhere near. That is the position, measured across 28 scored tasks. It is not a forecast about you.

So the thing worth your attention is not the job going away. It is the layer around it. Ordering medical and laboratory supplies and equipment is the part turning into software, and being the person who understands that layer is worth money.

This week: one thing

Ask the one question. Find whoever is bringing new software into your workplace (the manager, the office, whoever runs the system) and ask them what it is meant to do to medical and laboratory supplies and equipment, and what it is not meant to touch. Ten minutes, this week, before anyone decides it for you.

What you end up holding: a straight answer about what is actually being rolled out, and when
How long it takes: ten minutes

If there’s nobody obvious to ask, or you’d rather not ask your manager: Put the same question to your union rep, your shift lead or the person who has been there longest, in person, over a break. Same ten minutes, same answer, and you will usually get a straighter one. Write down what they say. The note is the artifact, and it tells you whether tentative diagnoses are in scope or not. Nothing to log into, no license needed.

Over the next 90 days

Get inside the tool rollout rather than waiting for it. Over the next ninety days, ask to be in the group that tests, checks or signs off whatever new system arrives near interpreting diagnostic test results for deviations from normal. It is usually an unglamorous seat that nobody fights for, and it is the one that decides how the software is used on your job rather than to it.

Over the next 12 months

On this evidence I would not retrain out of this job, and I will say that plainly rather than hedge it. The task list here is dominated by work that stays with a person. What I would do with a year is get formally recognised for the layer around it (the systems, the compliance, the planning), so you are the one who understands the software instead of the one it is done to. Before you pay for anything, use CareerOneStop - Find local training. It is free, it is the Labor Department's own service, and it is listed below with the rest of the free routes.

The roads out of here, and why I am not sending you down them

I looked at the obvious moves out of this job, and here is what I found.

I checked the 12 nearest US occupations to physician assistants (nearest by the work that AI is not taking, not by job title), and none of them survived. The closest was nurse anesthetists: only about 24% of its durable work is work you already do. And on the numbers you do not need one. This job scores 15/100 here, with only 2% of the task list in the top band, and “make tentative diagnoses and decisions about management and treatment of patients” is not work that hands over cleanly. None of them beats deepening what you already have.

How that was checked: this job was compared against all 830 US occupations in this release on their official task statements, and the 12 nearest were examined one by one. A move that turns on an industry, an employer or a qualification rather than on the work itself will not show up in a check like that. And this release carries no licence register, so anything you are weighing needs that looked up separately.

3 moves I checked and rejected

These are the obvious-looking jumps. They are here with their reasons rather than quietly dropped, because the ones that fail are worth knowing about. It is one less thing to turn over at night.

Nurse Anesthetists
Why it looked obvious: It came up as a near neighbour on the work AI is not taking: you already respond to emergency situations by providing cardiopulmonary resuscitation (CPR), basic cardiac life support…, and their equivalent is to respond to emergency situations by providing airway management, administering emergency fluids or drugs…. Across both published task lists that is about 24% of the durable work in that job.
Why I am not recommending it: You would be starting most of it from nothing: about 24% of the durable work in that job is work you do today, and the rest you would be learning while the bills carried on.
Look at that job’s page anyway →
Medical Assistants
Why it looked obvious: It came up as a near neighbour on the work AI is not taking: you already administer or order diagnostic tests, and their equivalent is to operate x-ray, electrocardiogram (EKG), or other equipment to administer routine diagnostic tests. Across both published task lists that is about 11% of the durable work in that job.
Why I am not recommending it: You would be starting most of it from nothing: about 11% of the durable work in that job is work you do today, and the rest you would be learning while the bills carried on. It is a pay cut, in those words: $45,690 against your $135,880, 66.4% less (OEWS May 2025 (both)). Retraining to earn less is a decision, not advice.
Look at that job’s page anyway →
Emergency Medicine Physicians
Why it looked obvious: It came up as a near neighbour on the work AI is not taking: you already administer or order diagnostic tests, and their equivalent is to select, request, perform, or interpret diagnostic procedures. Across both published task lists that is about 9% of the durable work in that job.
Why I am not recommending it: Almost none of it is work you already do: about 9% of the durable side of that job. That is a different job, not a next step. The pay gap is the market pricing a barrier: $335,550 against your $135,880 is 2.47× (OEWS May 2025 (both)), and you would be crossing it holding about 9% of their durable work. A gap that size with an overlap that small is a wish, not a route. And it is a narrow door: about 32,880 of those jobs against 162,150 of yours (OEWS May 2025), 20% as many seats.
Look at that job’s page anyway →

What I’d stop worrying about

A friend tells you what not to spend fear on. This is that list.

The headline number you read somewhere
The big “X% of jobs” figures are about the whole economy, not about you. The number that describes your job is on this page: 2% of its task weight, across 28 scored tasks. Every row behind it is printed above with the source. Check ours; ignore theirs.
The headlines about your trade disappearing
They are usually about the technology, not the timetable. Changes to work like examining patients to obtain information about their physical condition arrive through rules, insurance and money, slowly and visibly. This page tracks the task evidence and will move when it moves.
Retraining out of a job that is holding up
On this evidence I would not spend money leaving. Spend it on the layer around the job instead: the tools, the paperwork, the planning. That is where the change actually is.
The “obvious” next job everyone suggests
I checked the obvious moves and most of them did not survive. The reasons are printed with the routes above, including the pay and the gate. A move that fails on the numbers is worth knowing about so you can stop turning it over at night.

If you run a team doing this job

If you run a team doing this job, the conversation you owe them is the one on this page, and sooner than feels comfortable. Show them the ledger rather than a reassurance: the rows moving toward the software are ordering medical and laboratory supplies and equipment, and the rows that are not are where you want your people visible. Ask each of them to do the this-week move and bring the list to your next one-to-one. It turns a rumour into a piece of work, and it tells you which parts of your team's week are actually at stake. And say the thing out loud that a team lead usually leaves unsaid: a shrinking team is your exposure too, so do the move yourself as well.

You are reading the United States figures

The United Kingdom splits this work across more than one official group, of which Paramedics is the closest. The pay and employment figures are not directly comparable, and we do not average them together.

Switch to the United Kingdom page →partial match

In UK official statistics this job is counted as Paramedics. Pay and employment stay on this page’s own group; the task list and the scores do not cross over.

Your route through this

Where to go next, and what it costs

Free, and complete

The moves above cost nothing. These are the real services that go with them: public, government-funded, and free at the point of use. Nothing on this page is behind an email address or a payment.

Anywhere in the US:
CareerOneStop - Find local training
Search what's running near you, from the Labor Department's own database, before anyone sells you a course.
Free to search; individual programs vary, and some are funded
Anywhere in the US:
American Job Center locator
An American Job Center will sit down with you for free. Find yours by ZIP code.
Free
- CareerOneStop local help hub
Anywhere in the US:
CareerOneStop - Licensed occupations finder
Check what your state actually requires before you pay for anything.
Free
Anywhere in the US:
CareerOneStop
Free

Why there is no community here

Collab365, who build this site, run paid Spaces for a small number of subjects, and none of them is built for this job. We are not going to point you at the nearest one and call it a fit.

So the free services listed on this page are the whole answer, and it is the same answer we would give a friend.

Noted, and thank you. We’ll email you if a Space for physician assistants launches. Nothing else.

That did not look like an email address, so nothing was saved. Have another go below.

We could not save that. The fault is ours, not yours, and your address was not stored. Please try again later.

No Space for physician assistants yet. Should there be one?

Collab365 Spaces is built by the same people as this site. We find the problems that AI and automation are creating inside one kind of work, then solve them as short courses, briefings and Blueprints. Each Space is the community too, so the research and the people doing your job are in the same place.

What a Space actually is, in full

Collab365 launches new communities where the need is real. If one for physician assistants existed, with researched problems, courses and people in the same boat, would you want in?

Your email

We use your email address for one thing: to tell you if a Space for physician assistants launches. We never sell it, never use it for unrelated marketing, and every email has a one-click unsubscribe. Our privacy policy has the full version.

This unlocks nothing. Every figure, every row and every step on this page is already yours, whether you fill this in or not.

No deadline on any of this. The page will still be here, and the data is refreshed on a published schedule rather than when someone wants a headline.

Questions people ask about this job

Will AI replace Physician Assistants?: Not as a job, but it is already doing parts of the work. Across the 28 official task statements scored for Physician Assistants (United States, SOC 29-1071), 2% of the importance-weighted core work is made of tasks today's AI could already do most of. The overall exposure score is 15 out of 100 (range 12–20, band: minimal). That is a statement about tasks, not about headcount: this measures what AI could do, not whether any employer adopts it, whether the law allows it, or whether doing the routine parts faster creates more demand for the human parts. Figures are from release 2026-q4.1.
Which tasks in “Physician Assistants” can AI already do?: The highest-scoring tasks in release 2026-q4.1 are: “Order medical and laboratory supplies and equipment” (75/100, high); “Interpret diagnostic test results for deviations from normal” (47/100, partial); “Obtain, compile, and record patient medical data, including health history, progress notes, and results of physical examination” (42/100, partial). Each score comes from five published 0–4 ratings turned into a number by a published formula, and each carries the model's one-sentence reason on the page.
Which tasks in “Physician Assistants” stay human?: About 85% of this job's task weight sits in work that scores low for AI exposure. The lowest-scoring tasks in release 2026-q4.1 are: “Administer anesthetic, adjuvant, or accessory drugs under the direction of an anesthesiologist” (0/100, minimal); “Control anesthesia levels during procedures” (0/100, minimal); “Administer blood, blood products, or supportive fluids” (0/100, minimal). Low scores usually mean the task needs a body in a room, a legally accountable human, or trust built in real time. Those are the three things the scoring rubric treats as gates rather than obstacles.
What should someone working in “Physician Assistants” do about AI?: Start from the ledger rather than the headline: 2% of this job's weighted core work is exposed, and roughly 85% is not. The practical move is to spend more of your week on the tasks that score low, the ones above, and to get fluent at directing AI through the tasks that score high, because those are the parts that change whether or not you are ready for them. This page does not predict your job, and nothing here is career advice tailored to you: the score describes the occupation, not the person.
How is the AI exposure score for Physician Assistants calculated?: Each official task statement for the occupation is rated on five published 0–4 dimensions (output replicability, physical embodiment, licensed accountability, real-time human trust, and data availability) by claude-opus-5 using scoring prompt task_scoring_v1.0. The model never writes the score; a published formula turns the five ratings into a 0–100 number, so every score can be recomputed by hand. The occupation figure is the importance-weighted mean across 28 scored tasks. The prompt, the rubric, the formula and the full dataset are published at https://futureproof.collab365.com/method and https://futureproof.collab365.com/data/2026-q4.1 under CC BY 4.0.

Where these numbers come from

Worth knowing about these figures

The match between this job and its counterpart in the other country is partial, so the two pages count slightly different groups of people.
1 of this occupation's scored task statements carry a score that was measured under a different occupation's context, because the statement is shared between jobs and has only been scored once so far. Each one names the occupation it was measured under in the free routes below; none is presented as a measurement for this job.
Task clusters are not derived in this release, so the task-cluster field is empty and no Collab365 Space routing is attached to this occupation yet.

Task statements: onet-dbProcessing: catalogue-bridge → onet-im-rt-weighting → task-scoring → score-aggregation
Task weights: onet-db (im-rt)
Scores: Rubric task_scoring_v1.0, model claude-opus-5, computed 2026-08-04.
Pay and employment: bls-oews (May 2025 estimates (national_M2025_dl.xlsx))bls-oews (May 2025 estimates (national_M2025_dl.xlsx))

Figures on this page come from release 2026-q4.1, published 2026-08-05. Every release keeps its own permanent address, so a figure you cite in March is still there, unchanged, in November.

The plain-English wording on this page is assembled directly from the task statements and the published ratings, not written by hand for this occupation. That is why it is specific, and it is also why we say so.

The routes and free resources further up are today’s, not the release’s (last reviewed 2026-08-05). A route is an offer, not a historical fact, so it moves on its own clock.

How we score a job Download this release Look up another job

Using these figures?

Cite this

Everything on this site is published under CC BY 4.0. Quote it, chart it, sell something built on it. Just say where it came from, and cite the dated release rather than the site, so the figure you quote stays checkable.

Plain text

Collab365 (2026). Collab365 Futureproof: task-level AI exposure for US and UK occupations, release 2026-q4.1 (methodVersion 2.0.0, promptVersion task_scoring_v1.0). https://futureproof.collab365.com/data/2026-q4.1. Licensed CC BY 4.0. Built with O*NET data (USDOL/ETA, CC BY 4.0); ONS data (Open Government Licence v3.0); GAISI task framework (arXiv:2507.22748, MIT); BLS data (public domain).

BibTeX

@misc{collab365futureproof2026q41,
  title        = {Collab365 Futureproof: task-level AI exposure for US and UK occupations, release 2026-q4.1},
  author       = {{Collab365}},
  year         = {2026},
  url          = {https://futureproof.collab365.com/data/2026-q4.1},
  note         = {Release 2026-q4.1, methodVersion 2.0.0, promptVersion task_scoring_v1.0, CC BY 4.0}
}

Permalink

https://futureproof.collab365.com/data/2026-q4.1

Data as of release 2026-q4.1, published 2026-08-05. Releases never change after publication; when the figures move, a new dated release is published beside this one and this one stays exactly where it is.