Psychiatrists

29-1223.00
Median wage $281,870/yr27,980 employed (US)Rank #810 of 923 scored · top 88% by substitution

Diagnose, treat, and help prevent mental disorders.

Sub-scores

0–100 · band = confidence interval from rater disagreement

Substitution14
Exposure15
Augmentation52

Substitution — the headline: capability discounted by cost, barriers and adoption.

Exposure — technical capability alone, regardless of whether anyone deploys it.

Augmentation — how much AI assists without replacing. High here + moderate substitution = a changing job, not a disappearing one.

Tasks on the substitution scale

12 rated tasks, binned by substitution score.

Position among all scored occupations

Distribution of 923 occupation scores; the marker is this occupation.

Tasks with substitution ≥ 70

0%

Run 1.0.0-draft.1 · computed 2026-08-05 · rater panel: claude-sonnet-5, claude-haiku-4-5-20251001 · intervals span rater disagreement.

Why this score

The five weighted dimensions of the composite, averaged across this occupation's tasks (importance-weighted, panel mean). Exact weights and formulas: /api/v1/methodology.

Task automatabilityw 35%15

panel mean rating 1.6/5 → substitution pressure 15/100

Technical feasibility todayw 20%14

panel mean rating 1.5/5 → substitution pressure 14/100

Cost vs. human wagew 15%14

panel mean rating 1.6/5 → substitution pressure 14/100

Adoption barriersw 20%inverted — strong barriers lower the score9

panel mean rating 4.6/5 (barrier strength) → substitution pressure 9/100

Sector adoption velocityw 10%22

panel mean rating 1.9/5 → substitution pressure 22/100

Task breakdown (12 tasks)

Substitution pressure per task, weighted by O*NET importance in the composite. Expand a task for the full rater audit trail — every rating, every model, every rationale.

Prepare and submit case reports or summaries to government or mental health agencies.

31

CI 25–37 · exposure 33 · augmentation 63 · importance 3.1/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510012/5Psychiatry has lower digitization and AI integration rates compared to other medical specialties; adoption of AI-assisted documentation is still in early pilot phases within most mental health organizations, with conservative institutional attitudes toward automation of sensitive clinical records.
Sector adoption velocityclaude-sonnet-52/5Healthcare, especially mental health documentation to government agencies, has been slower to adopt AI tools compared to general professional services due to privacy, liability, and compliance concerns.
Augmentation potentialclaude-haiku-4-5-202510013/5AI can meaningfully assist psychiatrists by auto-populating structured fields, flagging missing information, and drafting initial summaries from clinical notes, improving documentation workflow. However, the assistance is limited to organization and drafting; clinical judgment and legal accountability remain fully human.
Augmentation potentialclaude-sonnet-54/5AI can meaningfully assist by drafting summaries from clinical notes, organizing case histories, and suggesting structure, saving significant physician time while they retain final review and responsibility.
Task automatabilityclaude-haiku-4-5-202510012/5AI can assist in drafting case reports by summarizing clinical notes and organizing information, but psychiatrists must interpret clinical judgment, synthesize complex diagnostic reasoning, and ensure legal/ethical compliance. End-to-end automation falls short of the 50% time-saving bar because human review and modification of sensitive mental health documentation is non-negotiable.
Task automatabilityclaude-sonnet-53/5Drafting structured case reports from clinical notes is amenable to AI summarization, but ensuring clinical accuracy, appropriate terminology, and compliance requires physician review, limiting full end-to-end automation.
Adoption barriersclaude-haiku-4-5-202510014/5Psychiatrists are legally and professionally accountable for case report accuracy and completeness submitted to government agencies. Mental health documentation carries privacy, liability, and regulatory requirements (HIPAA, state licensing boards) that create strong barriers to full automation and require licensed human verification and signature.
Adoption barriersclaude-sonnet-54/5These reports often carry legal and regulatory weight (e.g., commitment proceedings, insurance, licensing boards), requiring a licensed psychiatrist's certification and accountability for accuracy.
Cost vs. human wageclaude-haiku-4-5-202510012/5AI tools for medical documentation are moderately priced, but the psychiatrist's time spent reviewing, editing, and verifying AI-generated summaries often approaches the time of manual drafting. Oversight costs and the need for human sign-off limit cost advantage.
Cost vs. human wageclaude-sonnet-53/5AI drafting could cut time substantially, but psychiatrist review, correction, and sign-off remain necessary, keeping costs only moderately below fully manual preparation.
Technical feasibility todayclaude-haiku-4-5-202510012/5While NLP systems can extract and summarize medical data, no deployed product reliably generates psychiatry case reports meeting regulatory and clinical standards without substantial human oversight. Existing medical AI tools lack the contextual understanding and liability protection needed for mental health documentation submitted to government agencies.
Technical feasibility todayclaude-sonnet-52/5AI scribes and summarization tools exist in clinical settings but are not yet reliably deployed specifically for formal government/mental health agency case reporting with regulatory-grade accuracy.

Gather and maintain patient information and records, including social or medical history obtained from patients, relatives, or other professionals.

30

CI 28–32 · exposure 30 · augmentation 75 · importance 4.5/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510013/5EHR and documentation automation tools are in moderate use across psychiatric practices, but adoption of AI for patient interviewing and history-gathering remains limited. Many practices still rely on hybrid approaches (clinician + scribe) rather than fully AI-driven information collection.
Sector adoption velocityclaude-sonnet-53/5Healthcare is adopting AI scribes and documentation assistants at a moderate pace, with growing pilots in mental health but slower than fully digitized professional services.
Augmentation potentialclaude-haiku-4-5-202510014/5AI-powered transcription, structured data extraction from conversation, and EHR population tools significantly assist psychiatrists in documenting and organizing patient information, freeing time for analysis and clinical decision-making while the clinician remains responsible for conducting the actual interview and validating information accuracy.
Augmentation potentialclaude-sonnet-54/5AI ambient documentation and summarization tools meaningfully speed up recording and organizing patient history while the psychiatrist remains responsible for the actual clinical interaction.
Task automatabilityclaude-haiku-4-5-202510012/5While AI can assist with structured data entry and EHR documentation, gathering patient information requires nuanced clinical judgment, empathetic questioning, and contextual understanding that current systems cannot reliably perform end-to-end. The complexity of obtaining accurate social and medical history from multiple sources with varying reliability remains beyond full automation.
Task automatabilityclaude-sonnet-52/5AI can transcribe and organize information but gathering history requires interactive clinical interviewing, rapport-building, and judgment about what to probe, which current systems cannot fully replace end-to-end.
Adoption barriersclaude-haiku-4-5-202510014/5Psychiatry involves regulated healthcare delivery with strong medical-legal requirements for documentation accuracy and liability. Patient privacy (HIPAA), informed consent, and the requirement that a licensed psychiatrist assess and attest to clinical information create substantial barriers to full automation of this task.
Adoption barriersclaude-sonnet-54/5Clinical history-taking in psychiatry involves sensitive personal health information, licensure requirements, and liability concerns that require a qualified professional to conduct and verify the interview.
Cost vs. human wageclaude-haiku-4-5-202510012/5AI transcription and documentation assistance can reduce clerical time, but the psychiatrist's labor in the information-gathering interaction itself remains necessary and represents the bulk of the cost. AI tools reduce overhead but do not dramatically lower the per-task expense compared to the psychiatrist's loaded wage.
Cost vs. human wageclaude-sonnet-52/5AI scribe tools reduce documentation time cheaply, but the interviewing/history-gathering component still requires a licensed clinician's time, keeping overall cost comparable to human-only workflows.
Technical feasibility todayclaude-haiku-4-5-202510013/5Clinical documentation systems and EHR tools with AI-assisted transcription exist and are deployed, but they typically handle structured note-taking and data organization rather than the actual gathering of information from patients and collateral sources. Material gaps remain in accurately capturing nuance and synthesizing multi-source information.
Technical feasibility todayclaude-sonnet-52/5Ambient scribes and EHR summarization tools exist and are used in some clinics, but reliable end-to-end gathering of psychiatric history from multiple sources with clinical nuance is not yet a mature deployed capability.

Design individualized care plans, using a variety of treatments.

23

CI 20–25 · exposure 25 · augmentation 63 · importance 4.5/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510012/5Psychiatric care remains human-intensive with slow digitization; uptake of AI tools for treatment planning is in the pilot phase at most, with most psychiatrists still relying on manual planning and paper-based or simple EHR templates.
Sector adoption velocityclaude-sonnet-52/5Healthcare, especially psychiatry, has been slower than sectors like finance or tech to adopt AI for core clinical decision-making due to regulatory, ethical, and liability concerns.
Augmentation potentialclaude-haiku-4-5-202510013/5AI can usefully assist psychiatrists by summarizing patient histories, flagging drug interactions, suggesting evidence-based treatment options, and highlighting relevant clinical guidelines, raising clinician productivity in plan design while the psychiatrist retains final decision authority.
Augmentation potentialclaude-sonnet-54/5AI can meaningfully assist by summarizing patient history, suggesting evidence-based treatment options, and flagging interactions, enhancing psychiatrist efficiency while they retain final judgment.
Task automatabilityclaude-haiku-4-5-202510012/5While AI can assist in generating treatment suggestions and analyzing clinical data, designing individualized care plans requires integrating patient history, subtle clinical judgment, treatment interactions, and adaptive decision-making that current systems cannot reliably perform end-to-end at equal quality to a psychiatrist.
Task automatabilityclaude-sonnet-52/5Designing individualized psychiatric care plans requires integrating clinical judgment, patient history, risk assessment, and nuanced diagnostic reasoning that current AI cannot reliably replicate end-to-end.assistant
Adoption barriersclaude-haiku-4-5-202510014/5Regulatory frameworks (FDA, state medical boards) and liability law typically require that care plans be signed off by a licensed psychiatrist; malpractice risk is asymmetric (AI error can harm patients; psychiatrists remain liable), creating strong legal barriers to full automation.
Adoption barriersclaude-sonnet-55/5Psychiatric treatment planning is a licensed medical act requiring physician authorization, with high liability and legal requirements for clinician sign-off, making full automation legally barred.
Cost vs. human wageclaude-haiku-4-5-202510012/5AI-assisted clinical decision support is cost-comparable to or more expensive than clinician time when accounting for integration, curation of training data, liability risk, and mandatory oversight by a licensed psychiatrist.
Cost vs. human wageclaude-sonnet-52/5While AI-assisted drafting could reduce some documentation time, the physician's diagnostic and planning labor remains the dominant cost, so overall savings are modest compared to full automation scenarios.
Technical feasibility todayclaude-haiku-4-5-202510012/5No deployed AI product reliably designs complete individualized psychiatric care plans in production; decision-support tools exist but fall short of autonomous plan generation and require substantial clinician review and modification.
Technical feasibility todayclaude-sonnet-52/5Clinical decision-support tools exist that suggest treatment options or flag drug interactions, but no deployed product autonomously creates full individualized psychiatric treatment plans in production.

Analyze and evaluate patient data or test findings to diagnose nature or extent of mental disorder.

20

CI 20–20 · exposure 25 · augmentation 63 · importance 4.4/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510012/5Mental health organizations remain cautious adopters; psychiatry is a high-liability, heavily regulated domain with strong professional gatekeeping. Pilot use of AI for screening or data organization exists, but production replacement of diagnostic decision-making is rare and proceeding slowly.
Sector adoption velocityclaude-sonnet-52/5Healthcare, especially psychiatry, has historically slow AI adoption due to regulatory, liability, and trust barriers, with pilots more common than production deployment.
Augmentation potentialclaude-haiku-4-5-202510013/5AI can assist by organizing patient histories, flagging symptom patterns, and surfacing relevant differential diagnoses, moderately improving a psychiatrist's efficiency and diagnostic thoroughness. However, the human must retain full clinical authority, limiting the transformation potential compared to lower-stakes informational tasks.
Augmentation potentialclaude-sonnet-54/5AI can meaningfully assist by synthesizing patient history, flagging risk factors, summarizing test results, and suggesting differential diagnoses for physician review, improving efficiency while the psychiatrist retains final judgment.
Task automatabilityclaude-haiku-4-5-202510012/5While AI can assist in analyzing test data and identifying patterns in patient histories, diagnosing the nature and extent of mental disorder requires integrating subjective symptom reporting, contextual life factors, and clinical judgment that current systems cannot reliably perform end-to-end. The task demands synthesis of complex, often ambiguous psychological information that falls short of the 50% time-saving threshold for full automation.
Task automatabilityclaude-sonnet-52/5AI can assist with pattern recognition in test data and flag potential diagnoses, but the integrative clinical judgment required to diagnose mental disorders from complex, ambiguous patient data is not reliably automatable end-to-end today.“},
Adoption barriersclaude-haiku-4-5-202510015/5Legal and regulatory frameworks (medical licensing, liability standards, informed consent requirements) mandate that a licensed psychiatrist must perform or directly supervise diagnosis. This is a hard barrier: autonomous AI diagnosis of mental disorder would violate standard of care and licensing law in virtually all jurisdictions.
Adoption barriersclaude-sonnet-55/5Diagnosing mental disorders is a licensed medical act with strict legal, ethical, and liability requirements mandating physician judgment and sign-off.
Cost vs. human wageclaude-haiku-4-5-202510012/5AI inference costs are low, but integration, validation, and clinical oversight burden remain substantial. The liability and malpractice risk means organizations cannot reduce human psychiatrist involvement proportionally, keeping total cost-per-diagnosis near or above that of direct psychiatrist time.
Cost vs. human wageclaude-sonnet-52/5AI tools can cheaply process structured data or questionnaires, but since a licensed psychiatrist must still perform and validate the diagnosis, overall cost savings are limited by required human oversight.
Technical feasibility todayclaude-haiku-4-5-202510012/5Some products exist for symptom screening and data summarization (e.g., EHR-integrated analytics), but no deployed system reliably makes diagnoses independently. Clinical DSM-5/ICD-11 diagnosis requires nuanced interpretation of presentation, cultural context, and differential exclusion—areas where AI products show material error rates and narrow applicability in production.
Technical feasibility todayclaude-sonnet-52/5Some decision-support tools exist for screening (e.g., depression/anxiety scales, risk flagging) but no deployed product performs full diagnostic evaluation of mental disorders reliably in clinical production.

Teach, take continuing education classes, attend conferences or seminars, or conduct research and publish findings to increase understanding of mental, emotional, or behavioral states or disorders.

19

CI 7–30 · exposure 17 · augmentation 63 · importance 3.8/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510012/5While academic medicine is digitally mature, adoption of AI for autonomous research design, execution, and publication remains in pilot stages. Psychiatry as a field is cautious about AI replacing human scholarly and teaching roles; real displacement in this domain is minimal.
Sector adoption velocityclaude-sonnet-52/5Academic medicine and research adopt AI tools cautiously, with pilots for literature review and writing assistance but slow deployment for research design, teaching, and publication workflows.
Augmentation potentialclaude-haiku-4-5-202510013/5AI can usefully assist with literature review, data analysis, manuscript drafting, and identifying patterns in research datasets, raising a psychiatrist's productivity in preparation and analysis phases. However, the creative and interpretive core of research design and teaching still requires human leadership.
Augmentation potentialclaude-sonnet-54/5AI meaningfully assists with literature reviews, drafting manuscripts, preparing presentations, and synthesizing research findings, significantly aiding a psychiatrist's productivity in scholarly work.
Task automatabilityclaude-haiku-4-5-202510011/5This task requires original intellectual synthesis, novel research design, and teaching that demands human judgment, domain expertise, and the ability to interpret ambiguous clinical phenomena. AI cannot autonomously conduct rigorous psychiatric research, mentor trainees, or deliver credible continuing education at the level required in this field.
Task automatabilityclaude-sonnet-52/5AI can assist with literature review, drafting research papers, or summarizing conference content, but core activities like conducting original research, teaching interactively, and forming scholarly judgment require human expertise and cannot be automated end-to-end.
Adoption barriersclaude-haiku-4-5-202510015/5Publishing and teaching in psychiatry are protected by peer-review, institutional accreditation, and professional licensing. Only credentialed human experts can publish as principal investigators, lead research teams, and teach continuing education; regulatory and organizational structures mandate human accountability.
Adoption barriersclaude-sonnet-53/5No hard licensing requirement for research/teaching itself, but professional credibility, peer review, and academic norms create moderate friction against AI substitution, especially for original clinical research.
Cost vs. human wageclaude-haiku-4-5-202510011/5Conducting original psychiatric research, teaching, and publishing requires highly trained human psychiatrists whose loaded cost (salary, benefits, credentialing) is orders of magnitude higher than any AI tool, and AI cannot yet substitute for the intellectual and credibility demands of the task.
Cost vs. human wageclaude-sonnet-52/5AI tools reduce time spent on literature searches and drafting, but human oversight, data collection, clinical judgment, and publication processes still dominate cost, keeping savings modest relative to a psychiatrist's wage.
Technical feasibility todayclaude-haiku-4-5-202510012/5While AI can assist with literature review and draft generation of sections, no deployed system autonomously conducts psychiatric research, delivers clinical education, or publishes peer-reviewed findings. AI plays a supporting role in data analysis or writing, not an independent performer of the full task.
Technical feasibility todayclaude-sonnet-52/5AI writing and research assistants (e.g., literature summarization tools) are used narrowly for parts of this task, but no deployed product independently conducts psychiatric research or teaches with reliability at scale.

Review and evaluate treatment procedures and outcomes of other psychiatrists or medical professionals.

18

CI 15–20 · exposure 20 · augmentation 50 · importance 3.7/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510012/5Healthcare and psychiatric practice adoption of AI for clinical evaluation remains slow and cautious; most peer review continues to rely on manual processes by human clinicians, with AI primarily supporting documentation and data organization rather than independent clinical judgment.
Sector adoption velocityclaude-sonnet-52/5Healthcare peer review and quality assurance processes are historically slow to adopt automation due to regulatory, liability, and professional governance structures, despite AI adoption in adjacent clinical documentation tasks.
Augmentation potentialclaude-haiku-4-5-202510013/5AI can assist by extracting and summarizing treatment data, flagging statistical deviations, and organizing case information for review, moderately raising efficiency in the evidence-gathering phase while the psychiatrist retains full clinical evaluation responsibility.
Augmentation potentialclaude-sonnet-53/5AI can help summarize patient records, compare outcomes against clinical guidelines, and flag anomalies, meaningfully assisting the reviewing psychiatrist without replacing their judgment.
Task automatabilityclaude-haiku-4-5-202510012/5While AI can summarize treatment records and flag statistical outliers in outcomes, reviewing and evaluating peer clinical judgment requires nuanced assessment of complex cases, contextual reasoning, and professional accountability that current AI systems cannot reliably replicate end-to-end. Meaningful automation would require AI to substitute for human clinical expertise in ways that are not yet demonstrable.
Task automatabilityclaude-sonnet-52/5Peer review of treatment procedures requires clinical judgment, contextual reasoning about patient history, and accountability that current AI cannot autonomously replicate end-to-end, though it can assist with parts of chart review or literature checks.
Adoption barriersclaude-haiku-4-5-202510015/5Peer review and quality assurance in psychiatry are legally regulated functions typically requiring a licensed psychiatrist to conduct and attest to the evaluation; liability, credentialing, and professional responsibility frameworks create hard barriers to AI substitution.
Adoption barriersclaude-sonnet-55/5Peer review of medical treatment is typically legally and institutionally required to be conducted by licensed physicians (often psychiatrists) under credentialing, quality assurance, and malpractice liability frameworks, creating a hard barrier to AI substitution.
Cost vs. human wageclaude-haiku-4-5-202510012/5Current AI assistance in treatment review remains supplementary and requires psychiatrist oversight, making the all-in cost (AI + psychiatrist time) comparable to or higher than direct peer review by human psychiatrists without automation.
Cost vs. human wageclaude-sonnet-52/5AI tools could cheaply flag documentation gaps or outcome metrics, but the actual evaluative judgment still requires a licensed psychiatrist's oversight, keeping overall cost comparable to human-led review with modest efficiency gains.
Technical feasibility todayclaude-haiku-4-5-202510012/5No deployed product reliably performs independent peer review of psychiatric treatment decisions. AI tools exist for outcome tracking and documentation analysis, but evaluation of appropriateness and quality requires licensed clinical judgment that remains human-performed in production settings.
Technical feasibility todayclaude-sonnet-51/5No deployed product performs independent psychiatric peer review or quality assurance evaluation of colleagues' treatment decisions in production settings; this remains a human clinical governance function.

Examine or conduct laboratory or diagnostic tests on patients to provide information on general physical condition or mental disorder.

16

CI 11–21 · exposure 20 · augmentation 50 · importance 4.4/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510012/5Psychiatry remains largely traditional in diagnostic practice with slower digital transformation than fields like radiology; adoption of AI diagnostic tools in clinical psychiatry is still in pilot phases, not deep production deployment.
Sector adoption velocityclaude-sonnet-52/5Healthcare, especially psychiatric practice involving physical exams, has historically slow AI adoption due to regulatory, liability, and workflow integration barriers.
Augmentation potentialclaude-haiku-4-5-202510013/5AI can usefully flag abnormal lab values, suggest differential diagnoses from test results, and assist with documentation, but the psychiatrist retains responsibility for test selection, patient evaluation, and clinical synthesis, making this a moderate assistive role.
Augmentation potentialclaude-sonnet-53/5AI can assist with test result interpretation, flagging abnormal labs, and summarizing patient history, improving efficiency while the physician retains full responsibility for exam and diagnosis.
Task automatabilityclaude-haiku-4-5-202510012/5While AI can assist with interpretation of some diagnostic tests (e.g., analyzing lab results, EEG patterns), the task requires direct patient examination, clinical judgment about test selection, and integration with mental status assessment that current systems cannot perform end-to-end. AI cannot conduct the physical examination itself or make the nuanced decisions about which tests to order.
Task automatabilityclaude-sonnet-52/5AI can help interpret lab results or flag anomalies, but the hands-on examination and clinical judgment integrating physical exam with mental status assessment cannot be done end-to-end by current AI.",
Adoption barriersclaude-haiku-4-5-202510014/5Psychiatrists must legally order and interpret diagnostic tests; patient contact is required to justify test selection, obtain informed consent, and integrate results into clinical decision-making. Regulatory requirements (medical licensing, clinical judgment standards) create hard barriers to full automation.
Adoption barriersclaude-sonnet-55/5Ordering and interpreting diagnostic tests and conducting physical/mental exams requires a licensed physician, with strong liability, regulatory, and licensure requirements preventing automation.
Cost vs. human wageclaude-haiku-4-5-202510011/5Psychiatric diagnostic testing requires licensed clinician oversight, interpretation expertise, and regulatory compliance; the all-in cost of AI-assisted testing with required human review and liability coverage is not demonstrably cheaper than direct clinician ordering and interpretation.
Cost vs. human wageclaude-sonnet-51/5AI cannot substitute for the physical examination component, so no cost comparison favors AI for the core task; any AI use is supplementary, not a replacement of labor cost.
Technical feasibility todayclaude-haiku-4-5-202510012/5AI products exist for narrow diagnostic support (e.g., lab result interpretation, screening tools) but no deployed system reliably performs the full task of selecting, conducting, and contextualizing diagnostic tests within psychiatric evaluation. Human oversight remains mandatory, and error costs in mental health diagnostics are high.
Technical feasibility todayclaude-sonnet-51/5No deployed product performs physical examinations or administers diagnostic tests autonomously; this remains a physically embodied clinical task requiring a licensed physician.

Advise or inform guardians, relatives, or significant others of patients' conditions or treatment.

7

CI 0–15 · exposure 8 · augmentation 38 · importance 4.0/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510011/5Healthcare, particularly mental health, lags significantly in AI adoption for direct clinical communication. The combination of regulatory scrutiny, liability concerns, and the patient-sensitive nature of the work means adoption of AI for this task remains minimal.
Sector adoption velocityclaude-sonnet-52/5Healthcare, especially mental health, has been slow to adopt AI for direct patient/family-facing clinical communication due to regulatory, ethical, and trust concerns, with pilots more common than production use.
Augmentation potentialclaude-haiku-4-5-202510012/5AI could provide minor assistance (e.g., drafting templates for common conditions or organizing treatment summaries) but the task inherently requires the psychiatrist's judgment, empathy, and direct communication with family members, limiting meaningful augmentation.
Augmentation potentialclaude-sonnet-53/5AI can help psychiatrists prepare clear, well-organized explanations or answer general informational questions beforehand, but the human must still deliver and adapt the actual conversation.
Task automatabilityclaude-haiku-4-5-202510011/5This task requires complex interpersonal communication, emotional intelligence, and legal/ethical judgment about what information to share with whom. Current AI systems cannot reliably conduct the nuanced, real-time dialogue with family members that accounts for consent, cultural factors, and therapeutic context.
Task automatabilityclaude-sonnet-52/5This requires synthesizing sensitive clinical judgment, emotional nuance, and real-time responsiveness to family concerns, which current AI cannot reliably replace end-to-end despite being able to draft summary language.9 It also requires legal accountability and empathetic in-person communication that off-the-shelf systems cannot deliver.
Adoption barriersclaude-haiku-4-5-202510015/5Strong legal and regulatory barriers exist: psychiatrists must maintain confidentiality and informed consent, guardianship laws specify who can receive information, and professional licensing requirements mandate that a qualified psychiatrist perform or directly oversee this patient communication.
Adoption barriersclaude-sonnet-55/5Sharing patient information and clinical advice with family members is governed by strict privacy law (e.g., HIPAA), medical licensure, and liability requirements, mandating that a qualified psychiatrist perform or supervise this task.
Cost vs. human wageclaude-haiku-4-5-202510011/5The cost of integrating and overseeing an AI system to handle this sensitive communication, plus the cost of liability and regulatory non-compliance, far exceeds the cost of a psychiatrist spending time on it directly.
Cost vs. human wageclaude-sonnet-52/5While AI could cheaply draft informational summaries, the actual advising interaction requires a licensed clinician's time and liability coverage, keeping overall cost comparable to or greater than pure human delivery once oversight is included.
Technical feasibility todayclaude-haiku-4-5-202510011/5No deployed AI system reliably performs this task in production. While chatbots can provide general mental health information, the clinical and legal requirement for a licensed psychiatrist to communicate with guardians/relatives about specific patient conditions remains a standard of care with no AI replacement.
Technical feasibility todayclaude-sonnet-51/5No deployed product independently advises patients' families about psychiatric conditions or treatment in production; this remains firmly a clinician-delivered communication task.

Serve on committees to promote or maintain community mental health services or delivery systems.

4

CI 0–7 · exposure 0 · augmentation 38 · importance 3.0/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510011/5This task involves institutional governance and human decision-making. There is no meaningful AI adoption in committee service roles, nor would institutions substitute AI for required human expert participation.
Sector adoption velocityclaude-sonnet-52/5Healthcare administration and community health governance are slow-adopting sectors for AI in decision-making roles, with pilots focused on documentation rather than governance participation.
Augmentation potentialclaude-haiku-4-5-202510012/5AI could assist with background research, data synthesis, or preparation of briefing materials for committee meetings, but the core deliberative and representative functions require human psychiatrists to remain fully engaged.
Augmentation potentialclaude-sonnet-53/5AI can help prepare briefing materials, summarize community health data, draft meeting agendas, and analyze policy options, meaningfully aiding preparation even though it cannot replace the human's participatory role.
Task automatabilityclaude-haiku-4-5-202510011/5Committee service requires human judgment, consensus-building, and accountability for community mental health policy decisions. AI cannot meaningfully participate in deliberation, voting, or represent stakeholder interests in these governance bodies.
Task automatabilityclaude-sonnet-51/5This is a social, deliberative, relationship-driven governance activity requiring in-person judgment, negotiation, and community trust that current AI cannot perform end-to-end.rating remains minimal.
Adoption barriersclaude-haiku-4-5-202510015/5Legal and organizational requirements mandate that committee members be qualified professionals with formal accountability and fiduciary duty. Only licensed psychiatrists can serve in most formal mental health governance roles.
Adoption barriersclaude-sonnet-54/5Committee representation typically requires credentialed professional standing, institutional trust, and accountability that make substitution by non-human agents highly unlikely, though not legally mandated per se.
Cost vs. human wageclaude-haiku-4-5-202510011/5Committee service is a governance and strategic responsibility that does not have a direct alternative cost structure; a psychiatrist's time on a committee cannot be replaced by an AI system at any comparable price point.
Cost vs. human wageclaude-sonnet-51/5There is no AI product performing this role, so cost comparison favors the human by default; any AI use would only be a minor supplement, not a replacement of billable committee time.
Technical feasibility todayclaude-haiku-4-5-202510011/5No deployed product can autonomously serve on committees or participate in institutional governance. This task fundamentally requires human presence, accountability, and legal/fiduciary responsibility.
Technical feasibility todayclaude-sonnet-51/5No deployed product substitutes for a psychiatrist's participation in committee governance or policy deliberation; this is entirely outside current product scope.

Collaborate with physicians, psychologists, social workers, psychiatric nurses, or other professionals to discuss treatment plans and progress.

3

CI 0–6 · exposure 0 · augmentation 38 · importance 4.5/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510011/5Clinical collaboration is a core human-oversight requirement in mental health practice. Adoption of automation in this task is essentially non-existent; the work remains stubbornly interpersonal and professionally synchronous.
Sector adoption velocityclaude-sonnet-52/5Healthcare, especially psychiatric care, has been slower to adopt AI for core clinical collaboration tasks compared to sectors like finance or general information work, with adoption concentrated in administrative and documentation support rather than clinical judgment tasks.
Augmentation potentialclaude-haiku-4-5-202510012/5AI could assist in preparing summaries of prior notes or flagging relevant case history before a meeting, offering modest productivity gains in preparation. However, the core collaborative discussion itself is not substantially augmented by current AI.
Augmentation potentialclaude-sonnet-53/5AI can assist by summarizing patient records, drafting treatment plan documentation, or highlighting relevant clinical guidelines before or after these meetings, improving efficiency without replacing the collaborative decision-making itself.
Task automatabilityclaude-haiku-4-5-202510011/5Collaboration on treatment plans requires real-time interpersonal coordination, judgment of clinical context, and dynamic negotiation between professionals with different expertise. Current AI cannot meaningfully participate as a peer in these discussions or generate consensus across specialists.
Task automatabilityclaude-sonnet-51/5This is a real-time, multi-professional clinical collaboration requiring judgment, negotiation, and shared decision-making about patient care; no AI system can substitute for the psychiatrist's role in this interaction end-to-end.
Adoption barriersclaude-haiku-4-5-202510015/5Psychiatrists are legally responsible for treatment decisions and must document clinical reasoning with personal accountability. Regulatory and liability frameworks require the psychiatrist to own clinical judgments and collaborate as a licensed professional, not delegate to automated systems.
Adoption barriersclaude-sonnet-55/5Only licensed physicians can render psychiatric diagnoses and treatment decisions, and liability, medical licensing laws, and required clinical accountability make this collaboration non-delegable to AI.
Cost vs. human wageclaude-haiku-4-5-202510012/5AI tools (note summarization, meeting prep) offer modest cost reduction in preparation, but the core synchronous collaboration requires human psychiatrist time. The cost savings are minimal relative to the psychiatrist's loaded wage.
Cost vs. human wageclaude-sonnet-51/5There is no viable AI substitute performing this task, so cost comparison favors the human entirely; any AI role is only a minor supplement, not a replacement, so the cost ratio remains unfavorable to automation.
Technical feasibility todayclaude-haiku-4-5-202510011/5No deployed product reliably performs genuine multidisciplinary clinical collaboration. AI can summarize notes or suggest talking points, but cannot authentically participate in or lead treatment-plan discussions with licensed professionals.
Technical feasibility todayclaude-sonnet-51/5No deployed product conducts interdisciplinary treatment-planning collaboration in place of a psychiatrist; at most AI tools support note summarization or scheduling around such meetings.

Prescribe, direct, or administer psychotherapeutic treatments or medications to treat mental, emotional, or behavioral disorders.

1

CI 0–3 · exposure 0 · augmentation 50 · importance 4.8/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510011/5Psychiatry remains a heavily regulated, license-gated profession with high legal and ethical barriers to automation. Adoption of AI for treatment prescription is not occurring in production because it would violate law and professional standards; the sector is slow to displace core clinical decision-making.
Sector adoption velocityclaude-sonnet-52/5Healthcare, especially psychiatric prescribing, adopts AI slowly due to regulatory, liability, and safety constraints despite rapid AI adoption in some administrative healthcare functions.
Augmentation potentialclaude-haiku-4-5-202510013/5AI can assist psychiatrists with differential diagnosis support, medication interaction checking, and evidence-based treatment recommendations from literature, but the psychiatrist retains full clinical and legal responsibility for final treatment decisions. This represents useful but limited augmentation on a task fundamentally requiring human expert judgment.
Augmentation potentialclaude-sonnet-53/5AI can assist with diagnostic support, treatment guideline lookup, drug interaction checking, and documentation, meaningfully aiding psychiatrists without replacing their prescribing role.
Task automatabilityclaude-haiku-4-5-202510011/5Prescribing medications and directing psychotherapy require clinical judgment, patient assessment, diagnosis, and legal authority that current AI systems cannot perform end-to-end. Psychiatrists must evaluate complex patient histories, contraindications, and therapeutic response—tasks requiring licensed human oversight and accountability.
Task automatabilityclaude-sonnet-51/5This task requires clinical judgment, direct patient assessment, legal prescribing authority, and ongoing risk management that current AI cannot perform end-to-end.'
Adoption barriersclaude-haiku-4-5-202510015/5Strong hard barriers: psychiatric medication prescription and treatment direction require a state medical license and DEA registration. Liability, malpractice risk, and regulatory oversight of psychopharmacology are substantial, and mental health law in most jurisdictions mandates a licensed physician perform or legally sign off on all treatment decisions.
Adoption barriersclaude-sonnet-55/5Prescribing controlled substances and directing psychiatric treatment requires a licensed physician by law, with strict liability, DEA registration, and medical board oversight.
Cost vs. human wageclaude-haiku-4-5-202510011/5AI cannot reduce the cost of this task because the psychiatrist's time and legal liability cannot be substituted. Oversight and validation by the psychiatrist still dominate the cost structure, so the all-in cost to accomplish the outcome remains anchored to the human clinician.
Cost vs. human wageclaude-sonnet-51/5Since AI cannot legally or clinically substitute for the prescribing act itself, there is no viable AI-only cost comparison; any use requires physician oversight, adding cost rather than replacing it.
Technical feasibility todayclaude-haiku-4-5-202510011/5No deployed AI system can legally or reliably prescribe medications or direct psychotherapy independently. Clinical decision support tools exist, but they do not perform the core task of treatment prescription or therapeutic direction; a licensed psychiatrist must retain full responsibility.
Technical feasibility todayclaude-sonnet-51/5No deployed product independently prescribes or administers psychiatric treatment; AI tools at most support documentation or decision support alongside a physician.

Counsel outpatients or other patients during office visits.

1

CI 0–3 · exposure 0 · augmentation 38 · importance 4.3/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510011/5Psychiatry is a regulated medical discipline with strong professional and ethical norms against delegating patient counseling to non-human systems. Adoption of AI for direct counseling is effectively zero in practice.
Sector adoption velocityclaude-sonnet-52/5Healthcare, especially direct psychiatric care, is a slower-adopting sector for full task automation due to regulation, liability, and the need for human trust in mental health contexts.
Augmentation potentialclaude-haiku-4-5-202510012/5AI may assist with note-taking, appointment scheduling, or symptom tracking, but offers minimal augmentation to the core counseling act itself, which depends on clinical empathy, judgment, and therapeutic presence that AI cannot substantively enhance.
Augmentation potentialclaude-sonnet-53/5AI can assist with session notes, treatment plan drafting, symptom tracking, and even between-visit chatbot support, but it only marginally augments the actual counseling interaction itself.
Task automatabilityclaude-haiku-4-5-202510011/5Counseling requires nuanced emotional understanding, rapport-building, real-time clinical judgment, and therapeutic alliance—capabilities that current AI systems cannot reliably replicate. This task inherently demands human-to-human interaction and cannot meet the 50% time-saving bar with equal quality.
Task automatabilityclaude-sonnet-51/5Psychiatric counseling requires real-time clinical judgment, risk assessment (e.g., suicidality), rapport, and legal responsibility that current AI cannot replicate end-to-end; no off-the-shelf system substitutes for the office visit itself.
Adoption barriersclaude-haiku-4-5-202510015/5Psychiatrists must be licensed physicians with specific psychiatric credentials; they have legal and ethical duty to patients. Regulation and liability explicitly require a qualified human clinician to perform and sign off on counseling—a hard legal barrier.
Adoption barriersclaude-sonnet-55/5Psychiatric counseling requires a licensed physician, involves prescribing authority, liability for patient safety (e.g., suicide risk), and strict regulatory/ethical requirements for direct patient care.
Cost vs. human wageclaude-haiku-4-5-202510011/5AI systems capable of any meaningful counseling assistance remain expensive to develop and integrate, while a psychiatrist's billable time for counseling generates substantial revenue; the cost-benefit favors human provision.
Cost vs. human wageclaude-sonnet-51/5Because AI cannot legally or clinically replace the psychiatrist for this task, there is no viable cost comparison—human delivery remains mandatory, making AI substitution cost irrelevant/more expensive in practice.
Technical feasibility todayclaude-haiku-4-5-202510011/5No deployed AI product reliably performs psychiatric counseling in production. While chatbots can simulate conversation, they lack the clinical judgment, licensing, and therapeutic capacity required for actual patient care.
Technical feasibility todayclaude-sonnet-51/5No deployed product performs autonomous psychiatric counseling of outpatients in production; chatbot mental health apps exist but are adjunctive, not substitutes for licensed psychiatrist visits.

Related occupations — Healthcare Practitioners & Technical

How to read this

A high substitution score does not mean this job disappears — it means a large share of its current tasks face replacement pressure, so the mix of tasks is likely to change. High augmentation alongside substitution typically means the occupation reorganizes around the protected tasks. Wide confidence intervals mean the rater panel disagreed: treat those scores as open questions, not verdicts.

What would change this score

New model capabilities (automatability, feasibility), falling inference costs (cost ratio), regulation and licensing shifts (barriers), and measured sector adoption (velocity) all re-enter at every index release. Each release is recomputed, versioned and kept queryable — scores are claims with a date on them, not permanent labels.