AI exposure by occupation, O*NET 31-9092.00

Will AI replace Medical Assistants?

Medical Assistants: AI can already handle 1 of 20 tasks; the work mostly needs an efficient-tier model.

Have a team in this role?

Get your free 10-role sample: send this role and up to nine others, and a founder sends a board-pack PDF within two business days.

Every task, scored

The 20 tasks O*NET 30.3 lists for Medical Assistants, from most to least exposed. 1 reach 60 or more, the level at which AI already handles a task.

Task AI exposure Evidence Model tier Importance
Explain treatment procedures, medications, diets, or physicians' instructions to patients. 86 Observed Efficient 4.7
Interview patients to obtain medical information and measure their vital signs, weight, and height. 30 Estimated Efficient 4.8
Clean and sterilize instruments and dispose of contaminated supplies. 30 Estimated Efficient 4.8
Record patients' medical history, vital statistics, or information such as test results in medical records. 30 Estimated Efficient 4.7
Prepare treatment rooms for patient examinations, keeping the rooms neat and clean. 30 Estimated Efficient 4.7
Collect blood, tissue, or other laboratory specimens, log the specimens, and prepare them for testing. 30 Estimated Efficient 4.7
Show patients to examination rooms and prepare them for the physician. 30 Estimated Efficient 4.6
Perform routine laboratory tests and sample analyses. 30 Estimated Efficient 4.6
Greet and log in patients arriving at office or clinic. 30 Estimated Efficient 4.6
Perform general office duties, such as answering telephones, taking dictation, or completing insurance forms. 30 Estimated Efficient 4.5
Authorize drug refills and provide prescription information to pharmacies. 30 Estimated Standard 4.4
Change dressings on wounds. 30 Estimated Efficient 4.4
Schedule appointments for patients. 30 Estimated Efficient 4.2
Inventory and order medical, lab, or office supplies or equipment. 30 Estimated Standard 4.1
Contact medical facilities or departments to schedule patients for tests or admission. 30 Estimated Efficient 4.0
Operate x-ray, electrocardiogram (EKG), or other equipment to administer routine diagnostic tests. 30 Estimated Efficient 4.0
Set up medical laboratory equipment. 30 Estimated Efficient 3.7
Keep financial records or perform other bookkeeping duties, such as handling credit or collections or mailing monthly statements to patients. 30 Estimated Efficient 3.6
Help physicians examine and treat patients, handing them instruments or materials or performing such tasks as giving injections or removing sutures. 10 Estimated No tier 4.6
Prepare and administer medications as directed by a physician. 10 Estimated No tier 4.5

Observed: scored from AI use recorded in the Anthropic Economic Index (June 2026 release), 1 of 20 tasks. Estimated: scored from the task's O*NET 30.3 work activity, 19 of 20. Importance is O*NET's 1 to 5 rating of how central the task is. Model tier is the lowest tier that can do the task under the published rubric; "No tier" marks hands-on work.

The model tier this role needs

An efficient-tier model or lower covers 80% of the role's importance-weighted task work. 90% of the work is something AI can address at some tier, and 10% is hands-on.

Efficient, standard and frontier tiers come from the JobRoute model tier reference. The recommended tier is the lowest one that covers 80% of the work AI can address. How the method works.

Wages, employment and outlook

US employment
817,870
Median annual wage
$45,690
Middle half of earners
$38,470 to $49,180
Projected change, 2024 to 2034
+12.5%

Employment and wages: BLS Occupational Employment and Wage Statistics, May 2025, national. Projection: BLS Employment Projections 2024 to 2034.

Adjacent roles

The occupations that reuse the most of this role's skills, with the usual time to move. These are the first places to look when you plan to redeploy or retrain people in this role.

Skill overlap and transition time from the JobRoute role adjacency graph on O*NET 30.3 skills.

Questions about Medical Assistants and AI

Will AI replace Medical Assistants?

AI already handles 1 of the 20 tasks O*NET lists for this occupation, at a task exposure of 60 or more, and the role's importance-weighted exposure is 31 out of 100. JobRoute scores tasks, not whole jobs, so the score shows how much of the work changes rather than whether the job disappears. BLS projects employment to change by +12.5% from 2024 to 2034.

What is the AI exposure of Medical Assistants?

31 out of 100 on JobRoute's task exposure measure, the importance-weighted average of the role's O*NET task scores. Higher means more of the work is AI-doable today. 1 of the 20 task scores are observed in the Anthropic Economic Index (June 2026 release), and the other 19 are estimated from each task's O*NET work activity.

Which Medical Assistants tasks can AI already do?

The tasks with the highest exposure are: Explain treatment procedures, medications, diets, or physicians' instructions to patients. It is the only task that reach 60 or more.

What AI model tier does this role need?

An efficient-tier model or lower covers 80% of the role's importance-weighted task work. JobRoute assigns each task the lowest tier that can do it under a published rubric and recommends the lowest tier that covers 80% of the work AI can address.

How many people work as Medical Assistants, and what do they earn?

817,870 people in the United States, with a median annual wage of $45,690 (BLS Occupational Employment and Wage Statistics, May 2025). BLS Employment Projections expect a change of +12.5% from 2024 to 2034.

How to cite

Data version 2026-11-02, JobRoute method v1.0.0. Download all occupations or this occupation as JSON. Methodology.

JobRoute (2026). AI exposure: Medical Assistants. Data version 2026-11-02, method v1.0.0. Built on O*NET 30.3, BLS OEWS May 2025 and the Anthropic Economic Index (June 2026). CC BY 4.0. https://www.jobroute.ai/jobs/medical-assistants

Have a team in this role?

Get your free 10-role sample. Send this role and up to nine others, with no employee data, and a founder sends a PDF in board-pack format within two business days.

A founder replies within one business day.