SAFE
The documentation half of family medicine — visit notes, prior-auth letters, referral summaries, coding, patient-portal message drafts, guideline lookup — is already being absorbed by ambient scribes and clinical decision support, and that is real relief rather than real displacement. What remains is the part that pays: laying hands on a patient, examining an ear or an abdomen, doing skin biopsies and joint injections, deciding which of six vague complaints in a 15-minute slot is the dangerous one, and signing the prescription under a license you personally answer for. Longitudinal relationships with families over decades are the product, not a wrapper around it.
Roughly flat across the period, with year-to-year wobble.
Median pay $205,590 → $244,180 -5.0% in real terms
This line is counted by the Bureau of Labor Statistics — the one figure on this page that isn't a judgement of ours. Headcount moves on demand, offshoring, demographics and the business cycle, and automation is one term among several, often not the loudest.
So a falling line is not evidence that AI did it, and a rising one is not evidence that it won't. Both happen in this register: some occupations resist automation and shrink anyway, others are highly automatable and keep growing. The marked year is 2020.
BLS projection, 2024–2034
+2.7% 116,000 → 119,100 on the projections basis
Hard to automate, and growing
The work resists current AI and the BLS projects +2.7% more of these jobs by 2034. Note that safe does not mean well paid — several of the fastest-growing resistant occupations are among the lowest paid on the register.
Different clocks. The score is what current AI could do to this work today. The projection is how many of these jobs will exist in 2034. Everything between the two — how fast employers actually adopt, whether demand grows in the meantime — is why they can point opposite ways without either being wrong.
~3,300 openings a year on average, including replacing people who leave.
DoctorPhysicianGeriatricianFamily PhysicianGeneral PhysicianFamily PractitionerGeriatric PhysicianMedical Doctor (MD)General PractitionerOccupational PhysicianPrimary Care PhysicianMedical Staff PhysicianPublic Health PhysicianFamily Medicine PhysicianFamily Practice PractitionerBoard Certified Family PhysicianGeneral Practice Doctor (GP Doctor)Outpatient Family Medicine PhysicianFamily Practice Medical Doctor (FP MD)Family Practice Physician (FP Physician)Family Practice Physician Traditional PractitionerDO Physician (Doctor of Osteopathic Medicine Physician)Family Medicine PCP (Family Medicine Primary Care Physician)
Holding it up: liability shield . Weakest point: task resistance .
Mixed — a routine tier and a judgment tier Differential diagnosis from an undifferentiated complaint — the 15-minute slot that opens with 'I'm just tired' and has to end with a decision about whether this is anemia, depression, or occult malignancy — resists automation, but a genuinely large share of the workday is chart review, refill authorization, guideline-driven screening intervals, USPSTF-conformant health maintenance, and inbox triage that algorithms handle competently, which is what puts this at 13 rather than 17.
Hands-on in uncontrolled environments Otoscopy, abdominal palpation, pelvic and prostate exams, punch biopsies, cryotherapy, IUD insertion, joint injections and casting all require a physician's hands on a body in an exam room, but this is a controlled clinical environment with staff, equipment, and no roof to climb — hence 15 rather than the 19 of a wildland paramedic.
Licensed human required and personally liable State medical licensure, DEA registration for controlled substances, and board certification are hard legal gates; the prescription and the note carry your NPI, malpractice exposure attaches personally under state standard-of-care law, and no software can be the physician of record — this is the definitional 20.
Exists to be accountable for ambiguous calls You decide alone whether chest pain goes to the ED or home with a follow-up, whether to taper a long-term opioid patient, whether a child's fever justifies a lumbar puncture, and whether to file a mandated abuse report — high-stakes calls made on incomplete data with no specialist in the room, though referral to cardiology or surgery does distribute the hardest decisions, which is why 18 not 20.
The verdict above describes this occupation as a whole. Almost nobody does the typical version of a job — tick what's actually in your week and see how your own mix sits.
Your task mix speaks to task resistance (13/20 here) — how much of the day's work current AI already does. That is the dimension the boxes above are about.
It cannot move the other three. Liability shield (20/20) is whether the law requires a licensed human to sign. Trust premium (18/20) is whether buyers specifically pay for a person. Judgment and accountability (18/20) is whether the role exists to own consequential calls. Those are facts about the occupation's standing, not about which tasks are in your week — a paralegal who does only trial exhibits still holds no licence. Together they are 56 of this occupation's 84 points (67%).
Embodiment (15/20) is also a property of the work rather than the worker, but we don't tag individual tasks as physical or not, so the picker can't tell you anything about it. That's a limit of this tool, not a claim.
Did we get the list right? Tell us what's missing — the tasks are written from the outside, and you're reading this from the inside.
The moves above are yours to make. This is the other half: what would have to change in the world for the occupation itself to score higher. None of it is in any one person's gift, but it is where the floor actually comes from. Scores here are not a one-way ratchet. Only two of the five dimensions — task resistance and embodiment — track what machines can do. The other three track law, what buyers will pay for, and who is answerable, and those move in both directions, often in response to the same pressure AI creates. If every lever below landed, this occupation would score around 91/100, still SAFE.
Task-mix shift: as ambient scribes (Abridge, Nuance DAX) and inbox drafting absorb documentation and portal replies, the residual day concentrates into physical exam, in-office procedures (skin biopsy, joint injection, IUD placement, laceration repair), and undifferentiated-complaint triage — the tiers AI cannot close. Watch for procedure-heavy visit mix and panel size rising per physician without note-writing time.
Continued growth of direct primary care and concierge memberships (Atlas MD-style, ~$70-100/month) where the explicit product is a named physician's continuity — patients paying cash outside insurance for human relationship is the cleanest possible demonstration of a human-specific premium. Also watch payer recognition of longitudinal-relationship codes (G2211 add-on, live since 2024) which pays specifically for continuity.
If scope rules or malpractice carriers formalize the family physician as the accountable supervisor of AI-triaged and NP/PA-delivered panels — i.e. signing off on decisions made by others and by software across a larger panel — the role becomes almost purely consequential-call ownership. Watch state collaborative-practice agreements and carrier underwriting questions about AI supervision.
The limit. Already at 84 with liability_shield maxed; realistic headroom is a few points from task-mix concentration and continuity pricing. The real risk to this score is not AI but scope-of-practice expansion for NPs/PAs paired with AI decision support, which erodes the liability shield's exclusivity from the human side, not the machine side.
| New York-Newark-Jersey City, NY-NJ | 4,870 | $249,150 +2% |
| Boston-Cambridge-Newton, MA-NH | 2,950 | $284,010 +16% |
| Dallas-Fort Worth-Arlington, TX | 2,240 | $259,310 +6% |
| San Francisco-Oakland-Fremont, CA | 2,170 | $304,560 +25% |
| Atlanta-Sandy Springs-Roswell, GA | 2,060 | $259,320 +6% |
| Phoenix-Mesa-Chandler, AZ | 2,050 | $341,680 +40% |
| Detroit-Warren-Dearborn, MI | 1,940 | $181,990 -25% |
| Philadelphia-Camden-Wilmington, PA-NJ-DE-MD | 1,790 | $294,430 +21% |
| Oklahoma City, OK | 940 | $481,870 +97% |
| Knoxville, TN | 590 | $460,080 +88% |
| Great Falls, MT | 30 | $449,840 +84% |
Digital Health reports an AI scribe capability that transcribes GP telephone consultations and files them into patient records.
Has AI actually changed your work? One tap, anonymous, and the running tally is public. Nothing else is asked of you.
Rather than check back: get the digest and we'll tell you what changed — or watch a single occupation from its own page.