Psychiatrists are physicians who diagnose and treat mental illness with medication, psychotherapy and, when needed, hospital care. Chatbots that talk about feelings and software that screens for conditions make people ask whether AI could do the job, or whether psychiatry is a safe career to train for. This page answers with the official 2025 to 2035 projections from the US Bureau of Labor Statistics (BLS), O*NET task data, the FDA list of AI-enabled medical devices and the International Labour Organization (ILO) index of exposure to generative AI.
The short answer: not on current projections. The BLS projects psychiatrist jobs to grow 7.2 percent from 2025 to 2035, about twice the 3.5 percent for all jobs and twice the 3.6 percent for physicians overall. The reason it gives is demand: growing demand for psychiatric care and improved access to mental health services. It names no AI factor for psychiatrists. The ILO puts the group that includes psychiatrists in its lowest band, not exposed. The change to watch is a different one: nurse practitioners, who include psychiatric nurse practitioners, are projected to grow 41.0 percent and take on tasks physicians used to do.
What the BLS projects for psychiatrists and the mental health jobs around them
| Job (US SOC) | US jobs 2025 | Projected 2035 | Change | Openings a year | Median pay 2025 |
|---|---|---|---|---|---|
| Psychiatrists (29-1223) | 30,000 | 32,100 | +7.2% | 900 | $281,870 |
| Physicians (29-1210) | 808,800 | 837,900 | +3.6% | 20,800 | $271,090 |
| Nurse practitioners (29-1171), incl. psychiatric NPs | 336,300 | 474,100 | +41.0% | 29,400 | $132,300 |
| Clinical and counseling psychologists (19-3033) | 81,300 | 90,900 | +11.7% | 4,100 | $100,580 |
| School psychologists (19-3034) | 64,000 | 64,500 | +0.8% | 3,400 | $95,990 |
| Mental health and substance abuse counselors (21-1018) | 533,400 | 631,400 | +18.4% | 50,500 | $59,350 |
| Marriage and family therapists (21-1013) | 76,300 | 86,600 | +13.5% | 6,900 | $66,940 |
| Psychiatric technicians (29-2053) | 161,500 | 197,500 | +22.3% | 17,000 | $45,130 |
| All occupations | 170,280,800 | 176,198,500 | +3.5% | 17,461,800 | $50,980 |
Source: BLS Employment Projections, occupational projections and worker characteristics, 2025 to 2035. Median pay is the 2025 annual median. Psychiatrists are a small occupation, 30,000 jobs, with about 900 openings a year on average; most openings come from people who retire or leave, not from growth. Every mental health job in the table grows faster than all jobs except school psychologists.
Why the BLS expects more psychiatrists
The BLS Occupational Outlook Handbook says: "Growing demand for psychiatric care and improved access to mental health services will support job growth for psychiatrists." In its table of factors behind each projection, the BLS labels the psychiatrist factor a demand change: in private general hospitals, the share of psychiatrists "increases as growing demand for psychiatric services increases the need for psychiatrists" (BLS Table 1.12, factors affecting occupational utilization). That table names artificial intelligence as a factor for dozens of other occupations, such as customer service representatives and computer programmers; for psychiatrists it names none.
Where psychiatrists work, and where the jobs grow:
| Where psychiatrists work (NAICS) | Jobs 2025 | Share | Change to 2035 |
|---|---|---|---|
| Offices of physicians (621100) | 8,900 | 29.7% | +4.2% |
| General medical and surgical hospitals, private (62210P) | 4,000 | 13.4% | +7.9% |
| Outpatient mental health and substance abuse centers (621420) | 2,300 | 7.6% | +19.6% |
| Offices of mental health practitioners, except physicians (621330) | 1,900 | 6.2% | +17.0% |
| Psychiatric and substance abuse hospitals, state (62220S) | 1,700 | 5.6% | +3.3% |
| Educational services, state, local and private (colleges and universities) | 2,000 | 6.7% | +2.6% |
| State government, excluding education and hospitals | 1,600 | 5.3% | +1.9% |
| Local government, excluding education and hospitals | 1,200 | 4.1% | +3.9% |
| Residential mental health and substance abuse facilities (623220) | 300 | 1.1% | +9.8% |
| Self-employed | 900 | 3.2% | 0.0% |
| All industries | 30,000 | 100% | +7.2% |
Source: BLS National Employment Matrix, psychiatrists, 2025 to 2035. The fastest growth is outside the doctor's office: outpatient mental health and substance abuse centers (+19.6 percent) and offices of mental health practitioners (+17.0 percent), the same places where counselors, therapists and psychiatric technicians are projected to grow. Offices of physicians, the largest employer, grow 4.2 percent.
Will AI replace psychiatric nurse practitioners, or will they replace psychiatrists?
The BLS has no separate occupation for psychiatric nurse practitioners; they are counted with all nurse practitioners, projected to grow from 336,300 to 474,100 jobs (+41.0 percent), one of the fastest rates of any occupation. The BLS explains it as occupational substitution and demand change: "nurse practitioners are increasingly employed in team-based models of care, taking on tasks previously performed by physicians" (BLS Table 1.12), helped by expanding practice authority.
So the shift that the official data show in mental health care is between people, not from people to software: more prescribing and follow-up by nurse practitioners, alongside psychiatrists whose numbers still grow. For the nurse side, see will AI replace nurse practitioners?
Can AI diagnose mental illness?
Not on its own, as far as US regulators have cleared it. The FDA list of AI-enabled medical devices held 1,614 entries with final decisions from 1995 to June 2026. Most (1,230) are led by the radiology panel; none is filed under a psychiatry panel. The closest entries are six clearances for two autism diagnosis aids for young children, both filed under neurology:
- Canvas Dx (Cognoa) is "intended for use by healthcare providers as an aid in the diagnosis of Autism Spectrum Disorder (ASD)" for children aged 18 to 72 months, and "is not intended for use as a stand-alone diagnostic device but as an adjunct to the diagnostic process" (FDA 510(k) K243558).
- EarliPoint (EarliTec Diagnostics) is indicated "as a tool to aid qualified clinicians in the diagnosis and assessment" of autism in children aged 16 months to 7 years (FDA 510(k) K253442).
Both are aids that a clinician uses and signs off on. Diagnosis in psychiatry rests on talking with the patient and people around them, history, risk and judgment over time, and that is the part the data show staying with clinicians.
Where AI touches a psychiatrist's work
| Psychiatrist task (O*NET) | How AI tools touch it |
|---|---|
| Gather and maintain patient information and records, including social or medical history | The most exposed part: note drafting, record summaries and intake forms are text work that software already helps with |
| Analyze and evaluate patient data or test findings to diagnose the nature or extent of a mental disorder | Screening tools and diagnosis aids exist; the FDA-cleared ones are aids for a clinician, not stand-alone diagnosis |
| Prescribe, direct, or administer psychotherapeutic treatments or medications | Prescribing is a licensed act and stays with the physician or nurse practitioner, whatever software suggests |
| Counsel outpatients or other patients during office visits | Chatbots can offer support between visits, but the visit, the relationship and the clinical responsibility stay human |
| Collaborate with physicians, psychologists, social workers, psychiatric nurses or other professionals on treatment plans | Coordination and judgment across a care team; tools can share notes, not decide |
| Advise or inform guardians, relatives, or significant others of patients' conditions or treatment | Hard conversations with families, often about risk and consent, which stay with people |
Source: tasks from O*NET OnLine, Psychiatrists (29-1223.00). The right column is our reading, not a measured score. The pattern: records and paperwork are where software helps first, which could free time in a field where the BLS expects demand to grow; prescribing, risk decisions and the treatment relationship stay with licensed people.
How exposed psychiatry is to generative AI
The ILO Working Paper 140 (2025) scores each occupation's tasks from 0 to 1 for how much generative AI could do them, and groups occupations from not exposed up to gradient 4, the most exposed (how to read the ILO scores):
| Occupation (ISCO-08) | ILO band | Mean task score (0 to 1) |
|---|---|---|
| Psychologists (2634) | Gradient 1 | 0.39 |
| Generalist Medical Practitioners (2211) | Not exposed | 0.29 |
| Social Work and Counselling Professionals (2635) | Not exposed | 0.28 |
| Specialist Medical Practitioners (2212), the group that includes psychiatrists | Not exposed | 0.27 |
| Nursing Professionals (2221) | Not exposed | 0.25 |
| Nursing Associate Professionals (3221) | Minimal exposure | 0.22 |
Source: ILO 2025 index, ISCO-08 codes. Specialist doctors, the group that includes psychiatrists, score 0.27, in the not exposed band, below generalist doctors and counsellors. Psychologists score higher (0.39, gradient 1, the lowest exposed band), mainly through assessment and report writing; for that job and for counselling, see will AI replace therapists and psychologists? Clerical jobs such as data entry sit at gradient 4 (0.70). Exposure measures how much of the task list AI could help with, not how many jobs disappear.
What this means if you work in psychiatry or plan to
- Demand is the driver in the official data. Psychiatrist jobs grow 7.2 percent to 2035, twice the all-jobs rate, because of demand for mental health care; the BLS names no AI factor.
- The team is changing. Nurse practitioners grow 41.0 percent and take on tasks physicians used to do; counselors (+18.4 percent) and psychiatric technicians (+22.3 percent) grow too.
- Paperwork is the exposed part. Notes, records and screening are where AI tools help; cleared diagnostic AI is an aid for clinicians, not a replacement.
- Growth is in outpatient care. Outpatient mental health centers and mental health practices grow fastest; state psychiatric hospitals and government grow slowly.
For medicine as a whole, see will AI replace doctors? and will AI replace healthcare jobs? Projections are estimates, not guarantees, and they cover all causes of change, not AI alone.
Questions and answers
Will AI replace psychiatrists?
Not on current projections. The BLS projects psychiatrist jobs to grow 7.2 percent from 2025 to 2035, from 30,000 to 32,100, against 3.5 percent for all jobs, because of growing demand for psychiatric care and better access to mental health services. It names no AI factor for psychiatrists.
Is psychiatry safe from AI?
As safe as almost any job in the data. The ILO puts specialist doctors, the group that includes psychiatrists, in its not exposed band (0.27 on a 0 to 1 scale). AI tools help with notes, records and screening, while prescribing, risk decisions and treatment stay with licensed clinicians.
Will AI replace psychiatric nurse practitioners?
The data point the other way. The BLS counts psychiatric nurse practitioners with all nurse practitioners, projected to grow 41.0 percent from 2025 to 2035 as they take on tasks previously done by physicians in team-based care. It names no AI factor for them.
Can AI diagnose mental illness?
Not on its own in the US. The FDA list of AI-enabled devices (1,614 entries to June 2026) has no psychiatry panel entry. The closest are two autism diagnosis aids for young children, Canvas Dx and EarliPoint, cleared as aids for clinicians; Canvas Dx is not intended as a stand-alone diagnostic device.
How much do psychiatrists earn?
The BLS reports a median annual wage of $281,870 for psychiatrists in 2025, against $271,090 for physicians overall and $50,980 for all workers. It projects about 900 openings a year for psychiatrists from 2025 to 2035.