Physician Assistants
In short: today's best AI models could do about 38% of this job's working time (projected from METR's April 2026 measurement and Epoch's November 2026 index) if the work were set up for them, and about the same by the end of 2028 if progress keeps its long-run pace. The conversations, in-person and hands-on work, about 62% of the time, stays with people.
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.
162,150 people do this job in the US. $135,880 a year.
Reviewed on September 30, 2026: what each task is made of, and what handing it over would ask.
A typical 40-hour week
- AI could take on0 h
- A person checks AI's work15 h
- Desk work AI can't do yet0 h
- Stays with people25 h
Physician Assistants, now: of a 40-hour week, AI could take on 0 hours, 15 hours become checking its work, 0 hours are desk work it can't do yet, and 25 hours stay with people.
The largest pieces: Make diagnoses and treatment decisions (about 5 h, a person checks); Interpret test results (about 3 h, a person checks); Record patient history and notes (about 2 h, a person checks).
Estimates for a typical week of this job, if progress keeps its long-run pace, counting work the best models finish four times in five. Every task is below.
of the working time. Now is projected from METR's April 2026 measurement and Epoch's November 2026 index.
- 23%
- 15%
- 1%
- 23%
- 37%
of the work: Anthropic's measure of what people already bring to Claude at work (2025). It counts help with part of a task, so it can be higher than the share within reach.
On the chosen pace and reliability. Estimate.
Share of the working time within reach, now to 2030
Change assumptionsHide assumptionsLong-run pace, four in five
What each choice means: , , , , . Every pace starts from .
At this date and pace, models finish tasks up to 2.4 work days long four times in five.
What handing it over would take, now
The 38% within reach splits by what a model would need. The dates, pace and reliability follow the controls above. Estimates.
of the working time. The material can be handed to a model and the result handed back.
of the working time. The model has to be connected to the system the work lives in.
is conversation short enough by length. This site leaves conversation with people, so it is not in the total.
How far this has come
With GPT-4, measured in March 2023, 0% of this job's working time was within reach at . With Claude 3.7 Sonnet, measured in February 2025, 7%. With Claude Mythos Preview (early), measured in April 2026, 38%. Today's estimate: 38%.
A backcast: today's task data with the best model METR had measured by each date. It shows how fast the models have moved, not what anyone forecast at the time.
Measured use over time
Of the Claude.ai conversations Anthropic sampled and tied to a work task, 0.042% went to this job's tasks in May 2026, the 245th largest of the 447 jobs that showed up at all; 3 of its 12 tasks showed up. Within each of Anthropic's methods, never across a change: 0.016% to 0.038% (released February and March 2025); 0.096% to 0.065% (August 2025 to February 2026); 0.023% to 0.042% (April to May 2026).
Each release
| Conversations | Share | Rank | Tasks seen |
|---|---|---|---|
| Released February 2025 | 0.016% | 478 of 708 | 2 of 12 |
| Released March 2025 | 0.038% | 312 of 703 | 2 of 12 |
| The week of Aug 4, 2025 (released September 2025) | 0.096% | 161 of 655 | 5 of 12 |
| The week of Nov 13, 2025 (released January 2026) | 0.062% | 215 of 672 | 5 of 12 |
| The week of Feb 5, 2026 (released March 2026) | 0.065% | 214 of 677 | 4 of 12 |
| April 2026 (released June 2026) | 0.023% | 316 of 434 | 2 of 12 |
| May 2026 (released June 2026) | 0.042% | 245 of 447 | 3 of 12 |
Measured, not estimated: Anthropic's samples of Claude.ai conversations, each tied by a model to an O*NET task, from the Anthropic Economic Index. It says where Claude's use goes, not how much of this job's working time it covers, so it is not on the same scale as the reach above. The releases differ in their sample, their classifier and the smallest share they report (the June 2026 file rounds each task to a hundredth of a percent, so a smaller task reads as none), which is why a change of method starts a new line. At least 96% of each older release's task-tied use matched this site's tasks, and all of the newest. Claude.ai only, so use through other tools and other companies' models is not in it. Every job ranked.
Robots and the physical work
Anthropic rates 32% of this job's working time as physical work. For 0% of the working time, a robot has been shown doing most of the task somewhere. 0.00 of 3.
- Not physical 68%
- No robot shown 32%
Anthropic's rating, not this site's estimate: Claude Opus 5, searching the web, rated each physical O*NET task by the kind of setting in which a robot it could cite (deployed, sold or demonstrated) does most of the task, weighted by its own estimate of each task's share of the time (What work can robots do?, September 30, 2026, data CC BY 4.0; every job ranked). A robot shown somewhere is not a robot in your building: cost, setup and safety decide that. The share within reach on this page still counts no body work. The robot behind each rated task is under the task below.
What changes for the people in this job
From the task-by-task review, by the end of 2028 on the long-run pace at four in five. It says which parts of the work stay with people and which change, not whether anyone keeps a job: that depends on decisions no data set can see.
What stays with people
62% of the working time: the conversation, in-person and hands-on parts of the tasks.
- Assist in surgery, the hands-on work, 20%
- Counsel patients on care and health, the in-person work, 10%
- Examine patients, the hands-on work, 10%
- Perform therapeutic procedures, the hands-on work, 7%
- Make rounds and update charts, the in-person work, 4%
- Record patient history and notes, the in-person work, 4%
What becomes checking the machine
37% of the working time: desk work a model could do by then, in tasks where a mistake costs something or a licensed person signs. A person checks it. Another 2% is desk work that is small if wrong, which can be handed over with lighter checks.
- Make diagnoses and treatment decisions, a licensed person signs, 12%
- Interpret test results, a licensed person signs, 6%
- Record patient history and notes, severe if wrong, 6%
- Prescribe therapy or medication, a licensed person signs, 4%
- Order or administer diagnostic tests, a licensed person signs, 4%
- Supervise technicians, costly if wrong, 2%
What to learn
- Judgment where errors are severe. Some tasks are severe if wrong. Knowing when not to trust an output matters more there than speed.
- Signing off under your license. Some tasks need a licensed, certified or sworn person to do or sign them. The license, and the judgment it stands for, stays with the person.
- Reviewing work you did not do yourself. The tasks under What becomes checking the machine, where a mistake costs something, stay a person's to check or sign: spotting what a draft gets wrong is a skill of its own.
- The hands-on craft. About 37% of the time is hands-on work, which no model here does; robots are rated separately on this page.
- The conversations. About 24% of the time is conversation and in-person work, which this site leaves with people: listening, explaining, persuading, calming.
Your own version of this job: read it against your job description. The headline is free; the full personal report, every task with its timeline and what to learn, is $19 (a sample).
The tasks
Sorted by . 12 tasks from .
| Task | ||||
|---|---|---|---|---|
| Provide physicians with assistance during surgery or complicated medical procedures.Hands-on assistance during procedures.Done hourly or more by most who do it. Robots: no robot shown. Anthropic's rating. | 20% | 1.7 h25 min to 4.2 h | Hands-on 100% | Stays with people |
| Make tentative diagnoses and decisions about management and treatment of patients.Working out the diagnosis and plan from the findings.Done hourly or more by most who do it. | 12% | 21 min8 min to 50 min | Documents 100% | 100%Half by now |
| Instruct and counsel patients about prescribed therapeutic regimens, normal growth and development, family planning, emotional problems of daily living, and health maintenance.Instruction and counseling face to face.Done hourly or more by most who do it. | 10% | 21 min8 min to 38 min | In person 100% | Stays with people |
| Obtain, compile, and record patient medical data, including health history, progress notes, and results of physical examination.Taking the history from the patient and charting it.Done hourly or more by most who do it. | 10% | 17 min8 min to 38 min | Systems 60%, in person 40% | 60%Desk part half by now |
| Examine patients to obtain information about their physical condition.The physical examination, with questions as it goes.Done hourly or more by most who do it. Robots: no robot shown. Anthropic's rating. | 10% | 17 min8 min to 33 min | Hands-on 70%, in person 30% | Stays with people |
| Perform therapeutic procedures, such as injections, immunizations, suturing and wound care, and infection management.Injections, suturing and wound care by hand.Done several times a day by most who do it. Robots: no robot shown. Anthropic's rating. | 7% | 25 min8 min to 50 min | Hands-on 100% | Stays with people |
| Interpret diagnostic test results for deviations from normal.Reading results for what is abnormal.Done hourly or more by most who do it. AI observed on this task. | 6% | 13 min4 min to 25 min | Documents 100% | 100%Half by now |
| Visit and observe patients on hospital rounds or house calls, updating charts, ordering therapy, and reporting back to physician.Seeing patients on rounds, updating the chart and reporting to the physician.Done hourly or more by most who do it. Robots: no robot shown. Anthropic's rating. | 6% | 25 min13 min to 50 min | In person 50%, systems 30%, conversation 20% | 30%Desk part half by now |
| Administer or order diagnostic tests, such as x-ray, electrocardiogram, and laboratory tests.Orders entered in the record, with some tests run by hand.Done hourly or more by most who do it. | 6% | 13 min4 min to 25 min | Systems 60%, hands-on 40% | 60%Desk part half by now |
| Supervise and coordinate activities of technicians and technical assistants.Leading technicians in the clinic, with the planning behind it.Done daily by most who do it. | 5% | 25 min13 min to 50 min | In person 60%, documents 40% | 40%Desk part half by now |
| Prescribe therapy or medication with physician approval.Deciding the prescription and entering it for the pharmacy.Done hourly or more by most who do it. | 4% | 8 min4 min to 17 min | Documents 50%, systems 50% | 100%Half by now |
| Order medical and laboratory supplies and equipment.Checking the shelves, then placing the order with the supplier.Done more than yearly by most who do it. | 3% | 17 min8 min to 38 min | Systems 60%, hands-on 40% | 60%Desk part half by now |
Within reach, per task: the share of the task's time whose instances are short enough for a model to finish at the chosen date, pace and reliability. Half by: the date half of that time comes within reach. Made of: how the task's time splits among five kinds of doing. Only the documents and systems parts can count as within reach, so a task that is 30% documents can be at most 30% within reach.
Jobs that share skills with this one
From O*NET's related occupations. Within reach now and by the end of 2028, long-run pace, four in five.
Where this job works
The industry groups that employ the most of it (BLS, May 2025).
Where the most of these jobs are
The metro areas with the most people in it (BLS, May 2025).
Questions about this job
- How much of the work of Physician Assistants is within reach of AI now?
- An estimated 38% of the working time is within reach of AI models now and 38% by the end of 2028, on METR's long-run pace at four in five. Within reach means a model could finish work that long at that reliability. It is not a forecast of jobs lost or of savings.
- Which tasks of Physician Assistants come within reach first?
- Weighted by the time each takes, the tasks with the most within reach now (projected from METR's April 2026 measurement and Epoch's November 2026 index) are: Make diagnoses and treatment decisions (100% of the task's time now); Interpret test results (100% of the task's time now); and Record patient history and notes (60% of the task's time now).
- Which tasks of Physician Assistants stay with people?
- Tasks that are people work or body work are not counted as within reach at any date. The largest: Assist in surgery; Counsel patients on care and health; and Examine patients. Together such tasks are about 47% of the working time, and the people and body parts of the other tasks stay with people too.
- What should Physician Assistants learn as AI takes on more of the work?
- By the job's own task-by-task review, the first things to learn are judgment where errors are severe; signing off under your license; and reviewing work you did not do yourself. By the end of 2028 on the long-run pace at four in five, about 62% of the working time stays with people, 37% becomes checking a model's work and 2% could be handed over with lighter checks. It says which parts of the work change, never whether anyone keeps a job.
- Do people already use AI for the work of Physician Assistants?
- In the Claude.ai conversations Anthropic sampled for May 2026 and tied to a work task, 0.042% went to this job's tasks, the 245th largest share of the 447 jobs that showed up; 3 of its 12 tasks appeared. Anthropic's labor market measure puts the share of the job's tasks seen in real Claude use at work at 0%. These are measures of use, not of how much of the work AI could finish, and they cover Claude only.
- Can robots do the physical work of Physician Assistants?
- Anthropic's robot exposure rating (September 2026) finds 32% of the working time is physical work, and that a robot has been shown doing most of the task, somewhere, for 0% of the working time. Its index is 0.00 of 3. A robot shown somewhere is not one in every workplace; this site's own estimates still count no physical work as within reach.
- How is this estimated?
- From the 12 tasks O*NET lists for the occupation, each with an estimated length checked against GDPval and a share of the working time, reviewed one task at a time on September 30, 2026 for what each is made of (documents, business systems, conversation, in person, hands-on), and METR's measurements of how long a task the best AI models can finish, carried forward to the date from the latest, Claude Mythos Preview (early) in April 2026. Data checked November 10, 2026.
This job in your state or metro area
Embed these figures on your site
A small card with the share of working time within reach now and at the end of 2027, 2028 and 2030, the date and a link back. Free under CC BY 4.0; the card carries the credit.
Your own version of this job
The page above is the occupation. Your job is your title and what you actually do: give both, and they are read against these tasks in your browser, nothing sent anywhere. The headline is free: how much of your working time is within reach now and by the end of 2028, and how many of the tasks your description names.
The full personal report is $19, to print or save as a PDF: every task with when it comes within reach, which ones your description names, the parts of your description the task list does not cover, what stays with you, what becomes checking the machine and what to learn.