Allergists and Immunologists
29-1229.01Median wage $265,930/yr342,720 employed (US)Rank #730 of 923 scored · top 79% by substitution
Diagnose, treat, and help prevent allergic diseases and disease processes affecting the immune system.
Sub-scores
0–100 · band = confidence interval from rater disagreement
Substitution18
Exposure19
Augmentation62
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
16 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%17
panel mean rating 1.7/5 → substitution pressure 17/100
Technical feasibility todayw 20%22
panel mean rating 1.9/5 → substitution pressure 22/100
Cost vs. human wagew 15%18
panel mean rating 1.7/5 → substitution pressure 18/100
Adoption barriersw 20%inverted — strong barriers lower the score11
panel mean rating 4.5/5 (barrier strength) → substitution pressure 11/100
Sector adoption velocityw 10%25
panel mean rating 2.0/5 → substitution pressure 25/100
Task breakdown (16 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.
Document patients' medical histories.
69CI 67–70 · exposure 75 · augmentation 100 · importance 4.8/5 · click for rater detail
Document patients' medical histories.
69| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 4/5 | Healthcare is a digitized, information-intensive sector where EHR vendors have aggressively deployed AI documentation assistants. Adoption in hospital and ambulatory settings is rapid, with major health systems already piloting or deploying these tools in production. |
| Sector adoption velocity | claude-sonnet-5 | 3/5 | Ambient documentation tools are spreading quickly in larger health systems, but adoption in smaller specialty practices like allergy/immunology remains uneven and pilot-stage in many settings. |
| Augmentation potential | claude-haiku-4-5-20251001 | 5/5 | AI dramatically augments physician productivity by auto-populating histories, flagging inconsistencies, and generating structured summaries from voice or text input. Physicians remain in full control and oversight, making this a textbook example of high-value augmentation. |
| Augmentation potential | claude-sonnet-5 | 5/5 | AI scribing tools substantially reduce documentation burden and let physicians focus on patient interaction while remaining in the loop for review and finalization. |
| Task automatability | claude-haiku-4-5-20251001 | 4/5 | Current AI can extract and structure medical history information from unstructured notes, patient interviews, or prior records with high accuracy, achieving significant time savings. However, verification and clinical sign-off by a human remains necessary, preventing a full end-to-end rating. |
| Task automatability | claude-sonnet-5 | 4/5 | AI transcription and ambient-listening tools can capture and structure medical history documentation from patient encounters with substantial time savings, though physician review is still needed for accuracy and clinical nuance. |
| Adoption barriers | claude-haiku-4-5-20251001 | 3/5 | Regulatory requirements (HIPAA, medical record standards) and liability concerns around documentation accuracy create meaningful friction. However, no law requires a physician to personally type the history; AI-assisted and verified documentation is already legally accepted in most jurisdictions. |
| Adoption barriers | claude-sonnet-5 | 3/5 | Documentation itself isn't licensure-restricted, but physicians remain legally responsible for chart accuracy and must review AI-generated notes, creating moderate oversight friction and liability concerns. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 4/5 | AI-assisted documentation (via scribe tools and auto-population) costs substantially less than physician time spent manually entering history, though integration and oversight add overhead. The cost savings are typically 60–80% of manual physician effort. |
| Cost vs. human wage | claude-sonnet-5 | 4/5 | AI scribe subscriptions cost a small fraction of physician or scribe time per encounter, offering substantial savings even after accounting for review and correction overhead. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 4/5 | Deployed EHR systems with embedded NLP and medical record parsing reliably document routine histories at scale in healthcare organizations. Minor gaps remain in handling highly complex or non-standard cases, but production deployment is widespread. |
| Technical feasibility today | claude-sonnet-5 | 4/5 | Ambient clinical documentation products (e.g., DAX, Nuance, Abridge) are deployed in production across many health systems and reliably generate history notes from visits today, though not universal across all allergy/immunology practices. |
Educate patients about diagnoses, prognoses, or treatments.
30CI 29–31 · exposure 25 · augmentation 75 · importance 4.8/5 · click for rater detail
Educate patients about diagnoses, prognoses, or treatments.
30| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 3/5 | Healthcare is moderately digitized; many systems now use AI-powered patient portals and educational modules to supplement, but physician-delivered education remains the standard. Adoption is growing but largely in augmentative roles rather than full replacement. |
| Sector adoption velocity | claude-sonnet-5 | 2/5 | Healthcare, especially specialty medicine like allergy/immunology, adopts AI more cautiously than digital-native sectors, with patient education tools still largely supplementary. |
| Augmentation potential | claude-haiku-4-5-20251001 | 4/5 | AI can substantially assist by drafting personalized educational materials, summarizing complex concepts, generating visual aids, and identifying common patient questions—allowing physicians to deliver more comprehensive education in less time while maintaining the critical human relationship and judgment component. |
| Augmentation potential | claude-sonnet-5 | 4/5 | AI can help generate customized handouts, translate medical jargon, and prepare visit summaries, meaningfully aiding physicians in explaining conditions to patients even though the physician remains central. |
| Task automatability | claude-haiku-4-5-20251001 | 2/5 | While AI can generate educational content about allergies and immunology, the task requires personalized explanation tailored to individual patient circumstances, concerns, and comprehension levels—nuances that current AI struggles with reliably. End-to-end automation meeting the 50% time-saving bar is not yet demonstrated in production. |
| Task automatability | claude-sonnet-5 | 2/5 | AI can generate educational content and answer general questions, but personalized patient education tied to a specific diagnosis, prognosis, and treatment plan requires clinical judgment, interactive dialogue, and trust-building that current systems cannot fully replace end-to-end. |
| Adoption barriers | claude-haiku-4-5-20251001 | 4/5 | Regulatory requirements (FDA oversight of medical devices, state medical board rules), liability exposure for incorrect or misleading medical information, and the expectation that physicians personally educate and build trust with patients create substantial legal and professional barriers to full automation. |
| Adoption barriers | claude-sonnet-5 | 4/5 | Explaining diagnosis and treatment carries liability implications and is generally expected to be delivered or confirmed by a licensed physician, creating strong professional and legal barriers to full substitution. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 3/5 | AI-generated educational content and chatbots have low marginal cost per interaction, but require significant medical validation, legal review, and integration overhead to be safe and compliant. Fully loaded costs are roughly comparable to physician time for routine patient education. |
| Cost vs. human wage | claude-sonnet-5 | 3/5 | AI-generated educational materials are cheap to produce, but the physician's time for personalized counseling remains necessary, so overall cost savings are moderate rather than dramatic. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 2/5 | AI-driven patient education tools exist but typically serve as supplements (chatbots, informational websites) rather than replacements for physician-led education. Deployed systems lack the ability to deeply assess patient understanding, build rapport, and adjust explanations in real time as a clinician does. |
| Technical feasibility today | claude-sonnet-5 | 2/5 | Patient-facing chatbots and portals exist for general health information, but no deployed product reliably delivers personalized diagnosis/prognosis/treatment education in place of a physician at scale. |
Conduct laboratory or clinical research on allergy or immunology topics.
24CI 21–28 · exposure 25 · augmentation 75 · importance 3.6/5 · click for rater detail
Conduct laboratory or clinical research on allergy or immunology topics.
24| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 2/5 | While academic and biotech sectors are early adopters of AI analytical tools, actual research execution remains human-driven. Adoption of AI for core research conduct is slow; most usage is limited to data management and analysis assistance rather than autonomous research direction or execution. |
| Sector adoption velocity | claude-sonnet-5 | 3/5 | Biomedical research is adopting AI tools (data analysis, drug discovery, literature mining) at a moderate pace, with pilots and specific applications common but full research automation still emerging. |
| Augmentation potential | claude-haiku-4-5-20251001 | 4/5 | AI substantially augments researcher productivity through rapid literature synthesis, statistical analysis, data visualization, and protocol optimization—leaving the researcher in full control of hypothesis and experimental direction. These tools demonstrably accelerate research workflows and reduce time on routine analytical tasks. |
| Augmentation potential | claude-sonnet-5 | 4/5 | AI substantially augments research productivity via literature review, data analysis, experimental design suggestions, and pattern detection in immunological datasets, while researchers retain control over experimentation and interpretation. |
| Task automatability | claude-haiku-4-5-20251001 | 2/5 | Laboratory and clinical research involves complex experimental design, hypothesis formation, troubleshooting, and interpretation of results that require domain expertise and creative problem-solving. While AI can assist with literature review, data analysis, and protocol drafting, the core research direction and validation decisions remain fundamentally human tasks that AI cannot yet execute end-to-end at the quality level a researcher would achieve. |
| Task automatability | claude-sonnet-5 | 2/5 | AI can assist with literature review, data analysis, and hypothesis generation but cannot autonomously conduct laboratory experimentation, patient recruitment, or clinical research execution end-to-end.' |
| Adoption barriers | claude-haiku-4-5-20251001 | 4/5 | Research conduct, especially in clinical settings, is heavily regulated by IRBs, funding bodies, and institutional policies that require human researcher accountability and sign-off. Liability for research outcomes, data integrity, and ethical compliance creates substantial legal and organizational barriers to full automation or unsupervised AI conduct of research. |
| Adoption barriers | claude-sonnet-5 | 4/5 | Clinical and lab research involving human subjects requires IRB approval, licensed investigators, regulatory compliance (FDA/HIPAA), and physician oversight, creating strong institutional and legal barriers to full automation. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 1/5 | The cost of deploying AI systems for research oversight, validation, and integration would not approach the economies needed to displace a researcher; human expertise is still essential for directing the research itself, making the all-in cost substantially higher than the human labor it might theoretically supplement. |
| Cost vs. human wage | claude-sonnet-5 | 2/5 | AI reduces cost for literature synthesis and data analysis portions, but wet-lab work, clinical trial oversight, and specialized immunology expertise still require costly human researchers, keeping overall cost comparable to or only modestly below human-led research. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 2/5 | No deployed product reliably conducts original allergy or immunology research independently. AI tools exist for specific subtasks (literature mining, statistical analysis, data visualization) but genuine research conduct—including experiment design, hypothesis testing, and troubleshooting—remains a human-led activity with AI in supporting roles only. |
| Technical feasibility today | claude-sonnet-5 | 2/5 | AI tools (e.g., literature analysis, statistical software, lab automation for specific assays) exist and are used in research pipelines, but no product performs full research studies reliably without extensive human direction. |
Develop individualized treatment plans for patients, considering patient preferences, clinical data, or the risks and benefits of therapies.
24CI 20–28 · exposure 25 · augmentation 75 · importance 4.7/5 · click for rater detail
Develop individualized treatment plans for patients, considering patient preferences, clinical data, or the risks and benefits of therapies.
24| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 3/5 | Healthcare organizations are adopting clinical decision-support tools and AI-aided diagnostics at a moderate pace, with pilots common; however, autonomous treatment planning automation remains rare in production, and adoption is slower than in information and finance sectors due to regulatory and liability constraints. |
| Sector adoption velocity | claude-sonnet-5 | 2/5 | Healthcare, especially specialty medicine like allergy/immunology, has been slower to adopt AI for core treatment decisions compared to information/finance sectors, with pilots more common than production deployment. |
| Augmentation potential | claude-haiku-4-5-20251001 | 4/5 | AI can substantially assist clinicians by rapidly synthesizing literature, patient history, and risk-benefit evidence into structured option sets, reducing research time and supporting more thorough consideration of alternatives while the physician retains decision authority and personalization responsibility. |
| Augmentation potential | claude-sonnet-5 | 4/5 | AI can meaningfully assist by synthesizing clinical literature, flagging drug interactions, summarizing patient history, and suggesting evidence-based options, improving physician efficiency while they retain final judgment. |
| Task automatability | claude-haiku-4-5-20251001 | 2/5 | While AI can assist in synthesizing clinical data and generating evidence-based options, developing individualized treatment plans requires complex multi-factor judgment, patient preference integration, and accountability for risk-benefit trade-offs that current systems cannot reliably perform end-to-end without substantial human oversight and modification. |
| Task automatability | claude-sonnet-5 | 2/5 | Treatment planning requires integrating patient preferences, subjective risk tolerance, comorbidities, and nuanced clinical judgment that current AI cannot fully replicate end-to-end despite being able to summarize data or suggest options. |
| Adoption barriers | claude-haiku-4-5-20251001 | 4/5 | Physicians must legally evaluate and sign off on treatment plans; liability for adverse outcomes rests with the licensed clinician, creating a hard barrier to full automation and a strong requirement for human judgment and accountability that cannot be delegated to AI. |
| Adoption barriers | claude-sonnet-5 | 5/5 | Treatment plans must be authorized by a licensed physician, with legal, ethical, and regulatory requirements mandating physician sign-off, creating a hard barrier to full automation. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 2/5 | AI-based clinical decision support tools require significant integration, validation, and oversight costs alongside inference; combined with physician review time, the all-in cost per individualized plan approximates or exceeds the incremental cost of human clinician time. |
| Cost vs. human wage | claude-sonnet-5 | 2/5 | AI can cheaply generate draft suggestions, but the physician's diagnostic judgment, liability assumption, and patient discussion still require costly human time, keeping overall cost comparable to physician-led care. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 2/5 | AI products exist for clinical decision support and treatment recommendations, but they operate narrowly (e.g., specific conditions or drug interactions) and are not deployed as autonomous plan developers; physicians retain mandatory responsibility for plan validation, so no product reliably performs this task independently in production. |
| Technical feasibility today | claude-sonnet-5 | 2/5 | Clinical decision support tools exist and can suggest treatment options based on guidelines, but no deployed product autonomously develops finalized individualized treatment plans without physician authorship and oversight. |
Present research findings at national meetings or in peer-reviewed journals.
23CI 19–28 · exposure 17 · augmentation 75 · importance 3.3/5 · click for rater detail
Present research findings at national meetings or in peer-reviewed journals.
23| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 3/5 | Academic institutions and medical research are moderately digitized and cautiously experimenting with AI writing tools, but adoption remains slow and limited to drafting aids. Full automation of presentation authorship is rare in practice. |
| Sector adoption velocity | claude-sonnet-5 | 3/5 | Academic medicine has moderate AI adoption for writing assistance and literature review, but formal presentation and publication processes remain human-led with cautious institutional policies. |
| Augmentation potential | claude-haiku-4-5-20251001 | 4/5 | AI can meaningfully assist by drafting sections, organizing data visualizations, refining language, and generating outlines from raw findings, thereby accelerating the presentation-creation process while the researcher retains full control over content and claims. |
| Augmentation potential | claude-sonnet-5 | 4/5 | AI writing tools, citation managers, and slide generators meaningfully speed up manuscript drafting, literature synthesis, and presentation preparation for the physician-researcher. |
| Task automatability | claude-haiku-4-5-20251001 | 1/5 | Presenting research findings requires original intellectual contribution, interpretation of complex data, and rhetorical judgment tailored to a specific audience. Current AI cannot independently conceive novel findings or determine their significance—it can only assist in drafting or organization, not generate the core intellectual work. |
| Task automatability | claude-sonnet-5 | 2/5 | AI can help draft manuscripts, format abstracts, and generate slides, but the actual presentation, peer defense of findings, and authorship responsibility require human expertise and cannot be fully offloaded today. |
| Adoption barriers | claude-haiku-4-5-20251001 | 4/5 | Scientific journals and conference organizers have high editorial standards and require author accountability for novel claims. Professional reputation and peer credibility create strong human-contact and validation requirements that prevent full AI substitution. |
| Adoption barriers | claude-sonnet-5 | 4/5 | Journals and conferences require named, accountable authors with medical/scientific credentials, and academic integrity norms and journal policies restrict AI-generated content and authorship. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 2/5 | AI writing assistance is cheap, but the task demands expert human judgment that commands high opportunity cost. A trained allergist/immunologist's time spent on presentation far exceeds AI inference cost, making replacement economically unfavorable even with assistance. |
| Cost vs. human wage | claude-sonnet-5 | 2/5 | AI tools reduce drafting time cheaply, but the physician's time for study design interpretation, presentation delivery, and review response remains the dominant cost, limiting overall savings. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 2/5 | While AI can generate text and organize information, no deployed product can autonomously create novel research presentations that meet the scientific rigor and originality standards required for peer-reviewed journals or major conferences. Human researchers must validate and substantially revise any AI-generated content. |
| Technical feasibility today | claude-sonnet-5 | 2/5 | Writing assistants and slide-generation tools are used by researchers, but no deployed product reliably conducts the full research communication process including live Q&A and scientific judgment. |
Interpret diagnostic test results to make appropriate differential diagnoses.
21CI 16–25 · exposure 25 · augmentation 63 · importance 4.8/5 · click for rater detail
Interpret diagnostic test results to make appropriate differential diagnoses.
21| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 2/5 | Adoption remains slow and primarily in large medical centers as pilot projects; most allergists and immunologists rely on established human-driven workflows. Information-sector adoption patterns do not apply; healthcare remains risk-averse and regulatory-constrained, with deployment far behind research maturity. |
| Sector adoption velocity | claude-sonnet-5 | 2/5 | Healthcare overall adopts AI diagnostic aids slowly due to regulation, liability, and EHR integration challenges, with pilots more common than widespread production use in specialty diagnostics. |
| Augmentation potential | claude-haiku-4-5-20251001 | 3/5 | AI can usefully assist by highlighting relevant test patterns, flagging abnormal values, and suggesting differential diagnoses for review, raising clinician efficiency in the review process. However, the human physician must ultimately validate and integrate results, limiting the depth of augmentation compared to replacement-level automation. |
| Augmentation potential | claude-sonnet-5 | 4/5 | AI can meaningfully assist by summarizing lab data, cross-referencing symptoms with conditions, and suggesting possible differentials, improving physician efficiency while they retain final judgment. |
| Task automatability | claude-haiku-4-5-20251001 | 2/5 | While AI can assist in pattern recognition from test data, interpreting diagnostic results requires synthesis of clinical context, patient history, and nuanced judgment that AI cannot reliably execute end-to-end today. Current systems lack the integrated understanding needed to reliably suggest differential diagnoses that meet the 50% time-saving-at-equal-quality threshold without substantial human oversight. |
| Task automatability | claude-sonnet-5 | 2/5 | AI can flag abnormal values and suggest differentials but cannot reliably integrate patient history, exam findings, and test nuance to finalize a differential diagnosis without physician oversight. |
| Adoption barriers | claude-haiku-4-5-20251001 | 4/5 | Regulatory and liability barriers are substantial: diagnostic interpretation directly influences patient care decisions, and physicians face legal responsibility for diagnoses. Malpractice risk, regulatory scrutiny (FDA oversight of diagnostic AI), and the requirement that a licensed physician must evaluate and validate results create hard adoption friction. |
| Adoption barriers | claude-sonnet-5 | 5/5 | Diagnosis is a licensed medical act requiring physician sign-off, with high liability exposure, making this a hard regulatory and legal barrier to full automation. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 1/5 | AI diagnostic support systems require substantial domain-specific training data, specialized hardware, and continuous clinician oversight to validate outputs, making the total cost per task-equivalent comparable to or higher than a specialist immunologist's time, especially given liability and error costs. |
| Cost vs. human wage | claude-sonnet-5 | 2/5 | AI tools can cheaply generate suggestions, but the physician's required review, liability, and integration overhead keep overall costs comparable to or only modestly below physician time costs. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 2/5 | AI products exist for assisting with test interpretation (e.g., diagnostic decision support), but they operate with material error rates and narrow scope compared to human specialists. No deployed system reliably performs this task independently; systems are research-grade or limited pilot tools requiring heavy clinician validation. |
| Technical feasibility today | claude-sonnet-5 | 2/5 | Clinical decision-support tools exist that suggest differentials, but no deployed product autonomously interprets allergy/immunology test panels and reliably produces final diagnoses in production. |
Provide allergy or immunology consultation or education to physicians or other health care providers.
21CI 16–25 · exposure 17 · augmentation 63 · importance 3.9/5 · click for rater detail
Provide allergy or immunology consultation or education to physicians or other health care providers.
21| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 2/5 | Healthcare adoption of AI for clinical decision-support remains cautious and heavily gated by regulatory and liability concerns. Consultation roles in particular are slow to automate because they involve high-stakes advice and professional accountability structures that resist substitution. |
| Sector adoption velocity | claude-sonnet-5 | 2/5 | Healthcare specialist consultation remains a slower-adopting niche within medicine, with AI use concentrated in documentation and triage rather than specialist peer consultation. |
| Augmentation potential | claude-haiku-4-5-20251001 | 3/5 | AI can assist allergists and immunologists by rapidly retrieving and synthesizing clinical literature, flagging recent guideline updates, and drafting educational materials for peer consultation—meaningful productivity gains without replacing the specialist's judgment and authority. |
| Augmentation potential | claude-sonnet-5 | 4/5 | AI can strongly assist by summarizing literature, guidelines, and case data to support consultations, improving efficiency while the allergist/immunologist remains the authoritative decision-maker. |
| Task automatability | claude-haiku-4-5-20251001 | 2/5 | While AI can retrieve and summarize clinical literature on allergies and immunology, providing genuine consultation requires real-time clinical reasoning, contextual judgment about specific patient cases, and the ability to integrate complex medical histories—capabilities current AI systems lack reliably at production quality. Only narrow, pre-scripted educational components could be meaningfully automated. |
| Task automatability | claude-sonnet-5 | 1/5 | This requires an authorized specialist synthesizing complex immunology knowledge for peer consultation, integrating case-specific clinical judgment that AI cannot fully replace end-to-end today.' |
| Adoption barriers | claude-haiku-4-5-20251001 | 4/5 | Significant regulatory and professional barriers exist: medical consultation to other providers carries malpractice liability, clinical judgment authority typically requires a licensed physician, and healthcare organizations are highly risk-averse about substituting AI for expert medical advice. Physician oversight and sign-off are generally required. |
| Adoption barriers | claude-sonnet-5 | 4/5 | Delivering medical consultation and education to other providers typically requires licensure, credentialing, and professional liability accountability, creating strong structural barriers to full AI substitution. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 2/5 | An allergist-immunologist consultation commands significant professional fees; AI-generated summaries and educational content have minimal per-unit cost, but they cannot substitute for the specialized expertise being purchased. The cost comparison heavily favors the human because AI cannot deliver equivalent clinical value. |
| Cost vs. human wage | claude-sonnet-5 | 2/5 | Physician consultation time is costly, but the oversight, liability, and validation needed to trust AI-generated specialist advice add substantial hidden costs, keeping the ratio close to comparable. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 2/5 | No deployed product reliably performs expert-level immunology consultation to healthcare providers. AI can support literature search and generate educational summaries, but these fall short of the clinical judgment and liability-bearing consultation that characterizes the task. |
| Technical feasibility today | claude-sonnet-5 | 2/5 | AI tools (e.g., clinical decision support, medical LLMs) can provide reference information but no deployed product substitutes for specialist-to-specialist consultation in production clinical workflows. |
Assess the risks and benefits of therapies for allergic and immunologic disorders.
18CI 16–20 · exposure 25 · augmentation 75 · importance 4.5/5 · click for rater detail
Assess the risks and benefits of therapies for allergic and immunologic disorders.
18| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 2/5 | Healthcare remains a laggard sector for clinical-decision automation due to regulatory constraints, liability concerns, and organizational conservatism; AI adoption in allergy-immunology is primarily in diagnostic support, not therapeutic assessment. |
| Sector adoption velocity | claude-sonnet-5 | 2/5 | Healthcare, especially specialty medicine, adopts AI decision support cautiously due to regulatory, liability, and safety concerns, with pilots more common than full production deployment for this specific judgment task. |
| Augmentation potential | claude-haiku-4-5-20251001 | 4/5 | AI can substantially assist by rapidly aggregating clinical evidence, highlighting relevant studies, and organizing risk-benefit summaries, allowing physicians to focus on integrating findings with patient context and preferences—a clear productivity enhancer while the clinician retains judgment. |
| Augmentation potential | claude-sonnet-5 | 4/5 | AI can meaningfully assist by summarizing relevant literature, flagging drug interactions, and organizing patient data, significantly speeding up the physician's information-gathering process while they retain final judgment. |
| Task automatability | claude-haiku-4-5-20251001 | 2/5 | While AI can retrieve and summarize clinical trial data, synthesizing risk-benefit assessments requires clinical judgment that interprets nuanced patient contexts, comorbidities, and individualized risk tolerance—tasks current AI systems cannot reliably perform end-to-end at equal quality to physician assessment. |
| Task automatability | claude-sonnet-5 | 2/5 | Risk-benefit assessment for allergy/immunology therapies requires integrating patient-specific history, comorbidities, and nuanced clinical judgment that current AI cannot fully replicate end-to-end despite being able to summarize literature or flag interactions.' |
| Adoption barriers | claude-haiku-4-5-20251001 | 5/5 | Strong regulatory and legal barriers require a licensed physician to assess and take responsibility for therapy risk-benefit determinations; malpractice liability and standard-of-care expectations mandate human physician judgment and accountability. |
| Adoption barriers | claude-sonnet-5 | 5/5 | This is a core physician judgment task involving prescribing authority and liability; only a licensed allergist/immunologist can legally make and sign off on such risk-benefit determinations. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 1/5 | The integrated cost of AI tools for evidence synthesis plus required physician oversight and liability remains comparable to or higher than the cost of a physician performing the assessment directly. |
| Cost vs. human wage | claude-sonnet-5 | 2/5 | AI tools can cheaply generate draft summaries or flag risks, but the liability and complexity of immunologic therapy decisions still require expensive specialist physician time to finalize, keeping overall cost comparable to human-only processes. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 2/5 | AI tools exist for literature synthesis and evidence compilation, but no deployed product reliably performs independent risk-benefit assessment for clinical decision-making in immunotherapy; existing systems support clinicians rather than replace their judgment. |
| Technical feasibility today | claude-sonnet-5 | 2/5 | Clinical decision support tools exist that surface drug interactions and guideline recommendations, but no deployed product independently performs full risk-benefit assessments for allergy/immunology therapy without physician oversight. |
Engage in self-directed learning and continuing education activities.
17CI 11–23 · exposure 17 · augmentation 75 · importance 4.1/5 · click for rater detail
Engage in self-directed learning and continuing education activities.
17| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 2/5 | Physicians use AI tools to assist learning (literature review, content generation), but actual adoption of AI independently managing CME compliance is minimal, as regulatory and professional standards require the physician to own the learning process. |
| Sector adoption velocity | claude-sonnet-5 | 3/5 | Medical professionals increasingly use AI-assisted search and summarization tools for staying current, though formal CME systems remain traditional and slow to change. |
| Augmentation potential | claude-haiku-4-5-20251001 | 4/5 | AI significantly augments self-directed learning by generating summaries of recent literature, identifying relevant papers, creating study guides, and personalizing resource recommendations, materially raising a physician's efficiency in professional development activities. |
| Augmentation potential | claude-sonnet-5 | 4/5 | AI can significantly speed up literature review, summarize new studies, and personalize learning recommendations, meaningfully boosting the efficiency of self-directed learning. |
| Task automatability | claude-haiku-4-5-20251001 | 1/5 | Self-directed learning and continuing education require autonomous goal-setting, motivation, and adaptive learning choices based on evolving professional needs and knowledge gaps. Current AI systems cannot independently decide what to learn or pursue sustained educational development without human direction. |
| Task automatability | claude-sonnet-5 | 2/5 | AI can curate and summarize research and CME content, but the core act of self-directed learning is an internal cognitive process performed by the human, not something AI can execute end-to-end.' |
| Adoption barriers | claude-haiku-4-5-20251001 | 5/5 | Continuing medical education (CME) for board-certified allergists is often a legal/regulatory requirement that must be personally undertaken and documented by the licensed physician, not delegated to an AI system. |
| Adoption barriers | claude-sonnet-5 | 5/5 | Board certification and licensing bodies mandate that the licensed physician personally complete and attest to continuing education, making delegation to AI legally impossible. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 2/5 | AI can reduce the time spent gathering information or generating summaries, but the cost of integration and oversight for educational quality assurance is not negligible compared to the physician's own effort in selecting and consuming educational content. |
| Cost vs. human wage | claude-sonnet-5 | 2/5 | AI tools can cheaply supplement study materials, but since the task requires the physician's own comprehension and certification, there's no substitutable AI-only cost path. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 2/5 | While AI can generate study materials, summarize literature, or recommend resources, no deployed product demonstrates that an AI can independently engage in continuing education for a licensed physician. AI remains a tool that requires human initiation and judgment. |
| Technical feasibility today | claude-sonnet-5 | 2/5 | Products like literature summarizers and CME platforms exist but there is no deployed system that autonomously completes a physician's continuing education requirements for them. |
Prescribe medication such as antihistamines, antibiotics, and nasal, oral, topical, or inhaled glucocorticosteroids.
14CI 5–23 · exposure 17 · augmentation 63 · importance 4.8/5 · click for rater detail
Prescribe medication such as antihistamines, antibiotics, and nasal, oral, topical, or inhaled glucocorticosteroids.
14| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 1/5 | AI is not replacing prescribing; regulatory frameworks explicitly require human physicians to author and take responsibility for prescriptions. Adoption of AI for independent prescribing is effectively zero. |
| Sector adoption velocity | claude-sonnet-5 | 3/5 | Healthcare is adopting AI decision support and documentation tools at a moderate pace, with pilots for diagnostic and prescribing assistance but limited full-scale deployment for prescribing itself. |
| Augmentation potential | claude-haiku-4-5-20251001 | 3/5 | AI tools meaningfully assist prescribing through drug-interaction alerts, guideline summaries, and evidence lookup, allowing physicians to make better informed decisions faster. However, the physician remains the decision-maker and must evaluate AI suggestions critically. |
| Augmentation potential | claude-sonnet-5 | 4/5 | AI can meaningfully assist by suggesting drug interactions, dosing guidelines, and treatment protocols, improving physician efficiency and safety in prescribing decisions. |
| Task automatability | claude-haiku-4-5-20251001 | 1/5 | Prescribing medications is fundamentally a clinical decision-making task requiring diagnosis, patient history assessment, contraindication checks, and clinical judgment. Current AI cannot legally or reliably end-to-end prescribe medications independently; it lacks the diagnostic certainty and liability protection required. |
| Task automatability | claude-sonnet-5 | 2/5 | Prescribing requires clinical judgment integrating patient history, allergy testing, comorbidities, and legal accountability; AI can suggest options but cannot autonomously prescribe today.' |
| Adoption barriers | claude-haiku-4-5-20251001 | 5/5 | Legal and regulatory barriers are absolute: licensed physicians must write prescriptions in virtually all jurisdictions, and malpractice liability for adverse outcomes flows to the prescriber. Automation of this task is prohibited by law, not merely discouraged by practice. |
| Adoption barriers | claude-sonnet-5 | 5/5 | Prescribing medication is legally restricted to licensed physicians (or authorized prescribers), making this a hard regulatory barrier to full automation. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 1/5 | A physician's loaded wage for prescribing (including consultation, diagnosis, monitoring) remains far cheaper than building, integrating, validating, and maintaining AI systems that handle prescribing with appropriate liability and oversight. |
| Cost vs. human wage | claude-sonnet-5 | 2/5 | AI decision-support is cheap to run, but liability, oversight, and integration into EHR prescribing workflows keep effective cost comparable to physician time since a licensed prescriber must still review and sign. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 2/5 | While AI can assist with medication interaction checking and suggest treatment guidelines, no deployed product reliably prescribes medications independently in clinical practice. Products exist for decision support but humans must make and authorize the prescription. |
| Technical feasibility today | claude-sonnet-5 | 2/5 | Clinical decision support tools exist and can suggest medications, but no deployed product independently prescribes controlled or specialist medications in production without physician sign-off. |
Coordinate the care of patients with other health care professionals or support staff.
12CI 7–16 · exposure 9 · augmentation 63 · importance 4.3/5 · click for rater detail
Coordinate the care of patients with other health care professionals or support staff.
12| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 2/5 | While healthcare is investing in AI tools for scheduling and data management, autonomous care coordination by AI remains rare in production; adoption is limited to pilot phases with heavy human oversight. |
| Sector adoption velocity | claude-sonnet-5 | 2/5 | Healthcare adoption of AI for administrative support is growing but coordination of care remains largely manual due to complexity and liability concerns. |
| Augmentation potential | claude-haiku-4-5-20251001 | 3/5 | AI can usefully assist with scheduling, flagging communication gaps, and organizing patient data to streamline coordination, but the allergist or care coordinator remains the primary decision-maker and relationship manager. |
| Augmentation potential | claude-sonnet-5 | 4/5 | AI can meaningfully assist by summarizing records, drafting referral communications, and flagging care gaps, improving efficiency while the physician retains oversight. |
| Task automatability | claude-haiku-4-5-20251001 | 1/5 | Coordinating patient care with multiple healthcare professionals requires relationship-building, real-time decision-making, and interpersonal judgment that current AI systems cannot replicate end-to-end. While AI could assist in scheduling or documentation, it cannot autonomously manage the dynamic, context-dependent collaboration this task demands. |
| Task automatability | claude-sonnet-5 | 1/5 | Care coordination requires real-time judgment, negotiation, and relationship management across multiple clinicians that current AI cannot autonomously perform end-to-end. |
| Adoption barriers | claude-haiku-4-5-20251001 | 5/5 | Healthcare licensing laws, liability frameworks, and regulatory requirements (HIPAA, Joint Commission standards) mandate that licensed professionals retain responsibility for care coordination; legal and professional accountability cannot be delegated to AI systems. |
| Adoption barriers | claude-sonnet-5 | 4/5 | Clinical coordination involves licensed professional judgment, liability for care decisions, and regulatory/documentation requirements that necessitate human involvement. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 1/5 | Care coordination by an allergist or support staff involves real-time decision-making and accountability that AI cannot yet provide reliably; implementing oversight-intensive AI systems would likely exceed the cost of human coordination. |
| Cost vs. human wage | claude-sonnet-5 | 2/5 | AI tools can reduce administrative overhead but human oversight and communication remain necessary, keeping costs comparable rather than dramatically lower. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 2/5 | No deployed product reliably performs autonomous care coordination across multiple stakeholders; systems exist for scheduling and basic task routing but cannot handle the clinical judgment, accountability, and relationship dynamics required in real clinical settings. |
| Technical feasibility today | claude-sonnet-5 | 2/5 | Some products assist with scheduling, messaging, and summarizing records, but no deployed system autonomously coordinates multidisciplinary patient care reliably. |
Diagnose or treat allergic or immunologic conditions.
8CI 0–16 · exposure 13 · augmentation 63 · importance 4.8/5 · click for rater detail
Diagnose or treat allergic or immunologic conditions.
8| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 1/5 | Adoption of AI for independent diagnosis/treatment in allergology is minimal. These specialists work in regulated healthcare settings with strong professional and legal constraints on automation; pilots of decision-support tools are rare, and substitution of AI for clinical diagnosis is not occurring in practice. |
| Sector adoption velocity | claude-sonnet-5 | 2/5 | Healthcare overall adopts AI diagnostic/decision-support tools slowly due to regulatory, liability, and integration hurdles, with clinical decision-making still firmly human-led. |
| Augmentation potential | claude-haiku-4-5-20251001 | 3/5 | AI can meaningfully assist allergists by rapidly interpreting test panels, suggesting differential diagnoses, and flagging drug interactions or contraindications, reducing cognitive load and improving consistency. However, the human physician remains essential for history-taking, physical exam, patient communication, and final clinical judgment. |
| Augmentation potential | claude-sonnet-5 | 4/5 | AI can meaningfully assist allergists via symptom pattern analysis, differential diagnosis suggestions, and treatment guideline retrieval, improving efficiency while the physician retains final judgment. |
| Task automatability | claude-haiku-4-5-20251001 | 2/5 | Diagnosis of allergic/immunologic conditions requires integration of patient history, physical examination, specialized testing interpretation, and clinical judgment. Current AI can assist with pattern recognition in test results and differential diagnosis support, but cannot reliably conduct the full diagnostic workup or make treatment decisions autonomously without substantial human oversight and expertise. |
| Task automatability | claude-sonnet-5 | 1/5 | Diagnosing and treating allergic/immunologic conditions requires physical examination, testing, judgment about treatment plans, and legal medical authority; no current AI system can perform this end-to-end. |
| Adoption barriers | claude-haiku-4-5-20251001 | 5/5 | Legal and regulatory barriers are substantial: a licensed physician must clinically evaluate the patient, reach the diagnosis, and take responsibility for treatment decisions. Malpractice liability, medical licensing requirements, and the standard of care mandate human physician judgment; automation of the full task is prohibited by law and professional regulation. |
| Adoption barriers | claude-sonnet-5 | 5/5 | Medical licensure, malpractice liability, and regulatory requirements mandate a licensed physician to diagnose and treat patients, creating hard legal barriers to automation. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 1/5 | The cost of deploying AI for diagnostic support—including validation, regulatory compliance, integration with EHR systems, and required physician oversight—currently exceeds the cost savings from partial automation, given that allergists/immunologists command high billing rates and the liability implications of errors remain high. |
| Cost vs. human wage | claude-sonnet-5 | 1/5 | AI cannot independently perform this clinical task, so there is no valid cost comparison to a fully human-performed diagnosis-and-treatment cycle; human physician remains required. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 2/5 | While AI systems exist for symptom checking and test interpretation (e.g., image analysis for skin tests), no deployed product reliably performs end-to-end diagnosis or treatment planning for complex immunologic conditions in clinical practice. Existing tools have narrow scope and require physician validation; they do not meet the standard of mature production systems handling this task independently. |
| Technical feasibility today | claude-sonnet-5 | 1/5 | No deployed product autonomously diagnoses or treats patients for allergic/immunologic conditions; AI tools exist only as decision-support aids used by physicians. |
Order or perform diagnostic tests such as skin pricks and intradermal, patch, or delayed hypersensitivity tests.
4CI 0–7 · exposure 5 · augmentation 50 · importance 4.8/5 · click for rater detail
Order or perform diagnostic tests such as skin pricks and intradermal, patch, or delayed hypersensitivity tests.
4| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 1/5 | This task occurs in clinical settings where direct patient contact, physician oversight, and regulatory compliance are non-negotiable. Adoption of AI for autonomous test administration is not occurring because legal and safety barriers are absolute, not merely organizational. |
| Sector adoption velocity | claude-sonnet-5 | 2/5 | Healthcare adoption of AI for physical diagnostic procedures is slow due to regulatory, safety, and liability constraints, though AI is increasingly used for interpreting test results or triage. |
| Augmentation potential | claude-haiku-4-5-20251001 | 3/5 | AI can assist by automatically analyzing and reporting test results (e.g., measuring wheal diameter from standardized images, flagging unusual patterns), which can improve efficiency and documentation, but the human physician remains essential for patient assessment, test administration, and clinical decision-making. |
| Augmentation potential | claude-sonnet-5 | 3/5 | AI can assist in test result interpretation, generating differential diagnoses, or documenting findings after tests are physically administered, but does not aid the hands-on test performance itself. |
| Task automatability | claude-haiku-4-5-20251001 | 1/5 | Diagnostic skin prick and intradermal tests require precise physical manipulation of needles, real-time assessment of patient reactions, and clinical judgment that current AI cannot execute end-to-end. The physical dexterity, sterile technique, and need for immediate in-person adaptation make this task unsuitable for automation today. |
| Task automatability | claude-sonnet-5 | 1/5 | This task requires physically applying substances to a patient's skin, injecting allergens intradermally, and reading physical reactions over time—manual, hands-on procedures that current AI cannot perform end-to-end. |
| Adoption barriers | claude-haiku-4-5-20251001 | 5/5 | Medical licensure law explicitly requires a licensed physician or clinical immunologist to perform and interpret allergy testing; liability and regulatory requirements (FDA oversight of test devices, medical board regulations) create hard legal barriers to automation or delegation to non-licensed entities. |
| Adoption barriers | claude-sonnet-5 | 5/5 | Ordering and performing invasive diagnostic tests on patients requires licensed medical personnel, physical presence, and carries direct liability for adverse reactions like anaphylaxis, making this a hard-barrier clinical task. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 1/5 | Current AI has no viable end-to-end solution for performing these tests, making any cost comparison speculative. The human physician's loaded cost remains the baseline for actual test administration. |
| Cost vs. human wage | claude-sonnet-5 | 1/5 | AI cannot substitute for the physical test administration, so there is no viable AI cost comparison—human clinical staff must perform this regardless of AI availability. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 2/5 | While AI can assist in image analysis of test results (e.g., reading wheal sizes from photos), no deployed system performs the actual test administration or real-time clinical assessment autonomously. Existing products are limited to post-test analysis, not the core diagnostic procedure itself. |
| Technical feasibility today | claude-sonnet-5 | 1/5 | No deployed product performs skin prick, intradermal, patch, or delayed hypersensitivity testing; these remain manual clinical procedures performed by trained staff. |
Provide therapies, such as allergen immunotherapy or immunoglobin therapy, to treat immune conditions.
0CI 0–0 · exposure 0 · augmentation 38 · importance 4.7/5 · click for rater detail
Provide therapies, such as allergen immunotherapy or immunoglobin therapy, to treat immune conditions.
0| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 1/5 | Clinical immunotherapy administration remains a core physician responsibility with minimal automation adoption. The specialty is conservative, highly regulated, and patient-facing, limiting velocity of AI displacement in practice. |
| Sector adoption velocity | claude-sonnet-5 | 1/5 | Direct physical therapy administration in allergy/immunology clinics shows essentially no AI-driven displacement or adoption for this specific hands-on task. |
| Augmentation potential | claude-haiku-4-5-20251001 | 2/5 | AI could marginally assist with patient history analysis or literature review to inform therapy selection, but the core task of therapy delivery and monitoring remains fundamentally human-dependent with limited augmentation potential. |
| Augmentation potential | claude-sonnet-5 | 3/5 | AI can assist with dosing calculations, protocol selection, monitoring schedules, and documentation support around the therapy, though it doesn't touch the delivery itself. |
| Task automatability | claude-haiku-4-5-20251001 | 1/5 | This task requires hands-on clinical administration of specialized therapies (injections, infusions) and ongoing monitoring of patient immune responses. Current AI cannot physically administer treatments or make real-time clinical adjustments based on patient reactions. |
| Task automatability | claude-sonnet-5 | 1/5 | Administering injections, infusions, and managing therapy protocols requires physical presence, clinical judgment, and hands-on patient interaction that AI cannot perform. |
| Adoption barriers | claude-haiku-4-5-20251001 | 5/5 | Strong legal and regulatory barriers: only licensed physicians can prescribe and administer immunotherapy. Malpractice liability, patient contact requirements, and FDA oversight of immunological treatments create hard barriers to automation. |
| Adoption barriers | claude-sonnet-5 | 5/5 | Administering immunotherapy and immunoglobulin infusions requires licensed medical professionals, controlled substance handling, and legal liability for adverse reactions like anaphylaxis. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 1/5 | The task involves clinical decision-making, drug administration, and patient safety monitoring that demand a licensed allergist/immunologist. AI cannot replace this specialized labor cost, and any AI assistance would require expensive human oversight. |
| Cost vs. human wage | claude-sonnet-5 | 1/5 | AI has no mechanism to deliver physical therapies, so there is no substitutable AI cost basis; a human clinician remains mandatory. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 1/5 | No deployed AI system can independently administer immunotherapy or immunoglobulin therapy. These interventions require licensed physician oversight and direct patient contact, placing this firmly outside the scope of current automation products. |
| Technical feasibility today | claude-sonnet-5 | 1/5 | No deployed product administers immunotherapy or immunoglobulin therapy; this remains firmly a physician/nurse-delivered clinical procedure. |
Conduct physical examinations of patients.
0CI 0–0 · exposure 0 · augmentation 25 · importance 4.6/5 · click for rater detail
Conduct physical examinations of patients.
0| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 1/5 | Healthcare remains heavily regulated and conservative in automation; physical examination is a core clinical skill where human presence is legally and professionally mandated, preventing meaningful AI adoption despite sector digitization elsewhere. |
| Sector adoption velocity | claude-sonnet-5 | 1/5 | Direct physical examination remains almost entirely unautomated in clinical practice; healthcare delivery of hands-on exams shows minimal AI displacement despite AI adoption elsewhere in diagnostics. |
| Augmentation potential | claude-haiku-4-5-20251001 | 2/5 | AI can assist by suggesting differential diagnoses or flagging abnormal findings after the clinician completes the exam, but provides minimal real-time augmentation during the physical examination act itself. |
| Augmentation potential | claude-sonnet-5 | 2/5 | AI can support documentation (e.g., ambient scribes) or flag findings from data collected during exams, but offers little direct assistance to the physical act of examining a patient. |
| Task automatability | claude-haiku-4-5-20251001 | 1/5 | Physical examinations require direct human contact, tactile assessment (palpation, percussion), and real-time clinical observation that current AI cannot perform. While AI can assist in interpreting findings, it cannot conduct the examination itself. |
| Task automatability | claude-sonnet-5 | 1/5 | Physical examination requires hands-on palpation, auscultation, and observation of a patient's body, which current AI systems cannot perform without robotic embodiment that doesn't exist in deployed clinical practice. |
| Adoption barriers | claude-haiku-4-5-20251001 | 5/5 | Legal and regulatory requirements mandate that a licensed physician conduct patient physical examinations. Medical licensure, malpractice liability, and standard of care laws create hard barriers to substitution by non-human systems. |
| Adoption barriers | claude-sonnet-5 | 5/5 | Physical examination is a licensed medical act requiring direct physician-patient contact, with strong legal, regulatory, and liability requirements mandating human performance. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 1/5 | Physical examination automation is not economically viable today; the human clinician remains the only cost-effective method, and any attempted robotic substitute would be far more expensive than a physician conducting the exam directly. |
| Cost vs. human wage | claude-sonnet-5 | 1/5 | There is no AI substitute performing this physical task, so cost comparison favors the human by default since no alternative exists. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 1/5 | No deployed AI system performs end-to-end physical examinations. Telemedicine and diagnostic aids exist, but substituting the core examination—listening to lungs, palpating lymph nodes, checking vital signs—remains exclusively within human clinician domain. |
| Technical feasibility today | claude-sonnet-5 | 1/5 | No deployed product performs physical examinations autonomously; AI is at most used for interpreting images or data after a human has conducted the exam. |
Perform allergen provocation tests such as nasal, conjunctival, bronchial, oral, food, or medication challenges.
0CI 0–0 · exposure 0 · augmentation 25 · importance 4.2/5 · click for rater detail
Perform allergen provocation tests such as nasal, conjunctival, bronchial, oral, food, or medication challenges.
0| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 1/5 | Allergen provocation testing is a specialized clinical procedure that requires continuous human oversight and cannot be delegated to AI agents. There is minimal or no AI adoption in this domain, and regulatory and safety requirements make rapid AI substitution infeasible. |
| Sector adoption velocity | claude-sonnet-5 | 1/5 | Allergy/immunology procedural care is a low-digitization, hands-on clinical specialty with minimal AI displacement of physical testing procedures to date. |
| Augmentation potential | claude-haiku-4-5-20251001 | 2/5 | AI might assist with data logging, interpretation of past test results, or pre-test patient history synthesis, but the core task—administering allergen doses, monitoring live physiological responses, and making real-time safety decisions—cannot be substantially augmented by current AI systems. |
| Augmentation potential | claude-sonnet-5 | 2/5 | AI may assist with interpreting test results, tracking dosing protocols, or documentation, but offers little assistance during the actual physical challenge administration and monitoring. |
| Task automatability | claude-haiku-4-5-20251001 | 1/5 | Allergen provocation tests require direct physical administration of controlled allergen doses to patients, real-time monitoring of physiological responses, and immediate clinical judgment to stop tests if adverse reactions occur. Current AI cannot perform the hands-on clinical procedures or safely handle the unpredictable human response. |
| Task automatability | claude-sonnet-5 | 1/5 | This requires hands-on physical administration of allergens, real-time monitoring for anaphylaxis, and immediate clinical intervention capability that current AI cannot perform. |
| Adoption barriers | claude-haiku-4-5-20251001 | 5/5 | This task has hard legal barriers: only licensed allergists/immunologists can ethically and legally administer allergen challenges, manage acute reactions, and assume liability. Regulatory frameworks (FDA oversight of allergen materials, state medical practice acts) mandate human physician involvement. |
| Adoption barriers | claude-sonnet-5 | 5/5 | This is a high-risk medical procedure with anaphylaxis risk requiring licensed physician supervision, emergency protocols, and legal/liability requirements that mandate human execution. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 1/5 | An allergist performing this task commands a high professional wage, and the cost of automating the clinical infrastructure, monitoring equipment, and liability coverage would substantially exceed the cost of direct physician labor for a task requiring real-time clinical decision-making. |
| Cost vs. human wage | claude-sonnet-5 | 1/5 | AI cannot perform the physical test at all, so there is no substitutive cost comparison—human clinical staff and emergency equipment are mandatory. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 1/5 | No deployed AI system performs end-to-end allergen provocation testing. This task requires licensed physician presence, controlled clinical environments, and direct patient interaction that current AI tools cannot replicate in production. |
| Technical feasibility today | claude-sonnet-5 | 1/5 | No deployed AI product performs physical allergen challenge testing; this remains an entirely physician/nurse-administered procedure requiring direct patient contact and emergency readiness. |
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.