Physics Teachers, Postsecondary

25-1054.00

Median wage $100,310/yr13,090 employed (US)Rank #470 of 923 scored · top 51% by substitution

Teach courses pertaining to the laws of matter and energy. Includes both teachers primarily engaged in teaching and those who do a combination of teaching and research.

Sub-scores

0–100 · band = confidence interval from rater disagreement

Substitution28

Exposure25

Augmentation59

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

24 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

13%

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%25

panel mean rating 2.0/5 → substitution pressure 25/100

Technical feasibility todayw 20%25

panel mean rating 2.0/5 → substitution pressure 25/100

Cost vs. human wagew 15%31

panel mean rating 2.2/5 → substitution pressure 31/100

Adoption barriersw 20%inverted — strong barriers lower the score36

panel mean rating 3.6/5 (barrier strength) → substitution pressure 36/100

Sector adoption velocityw 10%25

panel mean rating 2.0/5 → substitution pressure 25/100

Task breakdown (24 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.

Maintain student attendance records, grades, and other required records.

CI 86–95 · exposure 92 · augmentation 63 · importance 4.1/5 · click for rater detail

Panel ratings per dimension for this task
Dimension	Model	Rating	Rationale
Sector adoption velocity	claude-haiku-4-5-20251001	5/5	Automated attendance and grade recording are already standard practice across postsecondary institutions, embedded in widely-used SIS platforms like Canvas, Blackboard, and Banner. This task has been substantially automated for over a decade in the higher education sector.
Sector adoption velocity	claude-sonnet-5	5/5	Higher education has near-universal adoption of digital gradebooks and LMS attendance systems already embedded in daily faculty workflow.
Augmentation potential	claude-haiku-4-5-20251001	3/5	AI assists faculty by automating data entry and consolidation, allowing instructors to focus on review and verification of records rather than manual input. While beneficial, the assistance is primarily on routine data handling rather than pedagogical or judgment-intensive aspects of the role.
Augmentation potential	claude-sonnet-5	4/5	Existing tools significantly reduce manual burden and error in tracking grades/attendance, though instructors still review and finalize records.
Task automatability	claude-haiku-4-5-20251001	4/5	Attendance tracking, grade recording, and routine record-keeping are highly structured, rule-based tasks that current AI can perform end-to-end with significant time savings. Integration with student information systems (SIS) is standard, and AI can handle data entry, verification, and organization with minimal human intervention, easily exceeding the 50% time-saving threshold.
Task automatability	claude-sonnet-5	5/5	Recording attendance and grades is a structured data-entry and calculation task easily handled by existing LMS/gradebook software with automation and AI-assisted grading tools, meeting the 50% time-saving bar.
Adoption barriers	claude-haiku-4-5-20251001	2/5	While FERPA and institutional data-governance policies create oversight requirements, there are no hard legal barriers preventing institutional automation of attendance and grade recording. Institutional friction and the need for human review of edge cases provide modest friction but not significant barriers to adoption.
Adoption barriers	claude-sonnet-5	2/5	Some institutional policies require instructor review/sign-off on final grades, but there is no licensing requirement preventing software from doing the record-keeping itself.
Cost vs. human wage	claude-haiku-4-5-20251001	5/5	The per-task cost of automated record maintenance through existing SIS infrastructure is orders of magnitude lower than paying a human to manually enter and organize student data. The marginal cost of processing additional records is negligible once the system is in place.
Cost vs. human wage	claude-sonnet-5	5/5	Automated record-keeping software costs a small institutional licensing fee compared to the faculty time it would take to manually maintain records, an order-of-magnitude cost reduction.
Technical feasibility today	claude-haiku-4-5-20251001	5/5	Educational SIS platforms with automated attendance logging, grade calculation, and record management are mature, widely deployed products in use across postsecondary institutions today. These systems reliably perform this task at scale with minimal error rates.
Technical feasibility today	claude-sonnet-5	5/5	Learning management systems (Canvas, Blackboard, PowerSchool, etc.) already perform automated attendance tracking and gradebook calculations reliably in production at scale across universities.

Compile bibliographies of specialized materials for outside reading assignments.

CI 76–92 · exposure 83 · augmentation 100 · importance 2.8/5 · click for rater detail

Panel ratings per dimension for this task
Dimension	Model	Rating	Rationale
Sector adoption velocity	claude-haiku-4-5-20251001	4/5	Academic institutions and postsecondary faculty have relatively high digitization and are early adopters of productivity tools. AI-assisted bibliography compilation is increasingly normalized in higher education, with widespread adoption in information-intense sectors.
Sector adoption velocity	claude-sonnet-5	3/5	Higher education is adopting AI tools for course prep and research assistance, but usage varies by institution and individual faculty, so it's not yet universal.
Augmentation potential	claude-haiku-4-5-20251001	5/5	AI dramatically assists faculty in this task by reducing search time, automatically formatting, and suggesting relevant materials, while the professor retains final judgment on pedagogical fit and reading list quality—a clear productivity multiplier.
Augmentation potential	claude-sonnet-5	5/5	AI substantially speeds up literature discovery and organization while the instructor retains final judgment over relevance and rigor of selected materials.
Task automatability	claude-haiku-4-5-20251001	4/5	AI systems can effectively search academic databases, identify relevant papers, and format bibliographies with high accuracy. The task is largely mechanical and benefits from AI's ability to process large volumes of structured information, though human curation of appropriateness remains valuable, representing a time-saving of well over 50% for compilation work.
Task automatability	claude-sonnet-5	5/5	AI can generate curated reading lists and bibliographies on specialized physics topics quickly, meeting or exceeding the time-saving threshold with equal or better quality when checked by the instructor.
Adoption barriers	claude-haiku-4-5-20251001	2/5	Few regulatory or legal barriers exist; institutional adoption mainly requires only that faculty trust the AI output quality. Light oversight (spot-checking sources) may be preferred but is not mandated, and the task does not require licensed credentials.
Adoption barriers	claude-sonnet-5	1/5	No licensing, regulatory, or human-contact requirement restricts using AI to compile reading lists; it's a low-stakes administrative/academic prep task.
Cost vs. human wage	claude-haiku-4-5-20251001	5/5	The cost of AI inference for bibliography compilation is negligible (cents per bibliography) compared to a professor's loaded hourly wage spent manually searching and formatting citations, creating at least a 10:1 cost advantage.
Cost vs. human wage	claude-sonnet-5	5/5	Generating a bibliography via an AI assistant costs a few cents in compute versus substantial faculty or TA time for manual literature searching and compiling.
Technical feasibility today	claude-haiku-4-5-20251001	4/5	Deployed AI tools including LLMs and academic search integrations reliably generate formatted bibliographies and can access major databases. Products like Zotero with AI plugins, ChatGPT with scholar search, and Perplexity perform this task in production, though occasional verification of sources remains best practice.
Technical feasibility today	claude-sonnet-5	4/5	Deployed tools like ChatGPT, research assistants, and citation managers reliably compile topical bibliographies today, though occasional citation inaccuracies require verification.

Prepare course materials, such as syllabi, homework assignments, and handouts.

CI 76–81 · exposure 75 · augmentation 100 · importance 4.5/5 · click for rater detail

Panel ratings per dimension for this task
Dimension	Model	Rating	Rationale
Sector adoption velocity	claude-haiku-4-5-20251001	3/5	Adoption is growing in higher education but remains patchy; pilots and early adopters are common, but systematic, institution-wide displacement of this prep work remains limited and varies by discipline and institution maturity.
Sector adoption velocity	claude-sonnet-5	3/5	Higher education has moderate AI adoption for content creation, with growing use but institutional caution and inconsistent policies across departments.
Augmentation potential	claude-haiku-4-5-20251001	5/5	AI transforms productivity on this task: instructors use AI to draft and iterate materials rapidly, customize for their pedagogy, and generate variations (alternative problem sets, accommodations), keeping the instructor firmly in control and judgment.
Augmentation potential	claude-sonnet-5	5/5	AI substantially speeds up drafting of syllabi, problem sets, and handouts while the instructor retains control over accuracy, pedagogy, and final content choices.
Task automatability	claude-haiku-4-5-20251001	4/5	AI can generate syllabi, homework assignments, and handouts at high quality with minimal human oversight, easily meeting the 50% time-saving threshold. Current LLMs handle content generation, formatting, and customization well, though human review for institutional compliance and pedagogical fit remains important.
Task automatability	claude-sonnet-5	4/5	LLMs can draft syllabi, homework sets, and handouts covering standard physics topics quickly, requiring mostly review and light editing rather than creation from scratch.4
Adoption barriers	claude-haiku-4-5-20251001	2/5	No legal or licensing barrier requires a human to write syllabi or homework; institutional review processes and teaching philosophy preferences create modest friction but not hard blocks to automation.
Adoption barriers	claude-sonnet-5	1/5	No licensing or regulatory requirement mandates a human create these materials; institutional norms allow instructors to use any drafting tool they choose.
Cost vs. human wage	claude-haiku-4-5-20251001	5/5	AI inference cost for generating a semester's materials is negligible (cents to low dollars) compared to a professor's hourly wage for the same task, yielding orders-of-magnitude savings in direct task execution cost.
Cost vs. human wage	claude-sonnet-5	5/5	Generating drafts of course materials via AI costs cents per document versus substantial faculty or TA time, an order-of-magnitude cost reduction.
Technical feasibility today	claude-haiku-4-5-20251001	4/5	Mature AI tools (ChatGPT, Claude, specialized edu-tech platforms) demonstrably produce functional course materials in production use by educators. Performance is reliable for routine generation, though edge cases around specialized physics content or complex problem design may require iteration.
Technical feasibility today	claude-sonnet-5	4/5	Widely deployed tools like ChatGPT, Copilot, and course-design AI assistants are already used by instructors to generate syllabi and assignments, though instructor review for accuracy is standard practice.

Participate in student recruitment, registration, and placement activities.

CI 30–87 · exposure 58 · augmentation 63 · importance 3.1/5 · click for rater detail

Panel ratings per dimension for this task
Dimension	Model	Rating	Rationale
Sector adoption velocity	claude-haiku-4-5-20251001	4/5	Higher-education institutions have rapidly adopted CRM platforms, chatbots, and automated application systems for recruitment and registration, with widespread production deployments in mid-to-large universities.
Sector adoption velocity	claude-sonnet-5	2/5	Higher education administrative functions have been slow to adopt AI agents for recruitment beyond basic chatbots and CRM tools; faculty-level involvement remains largely manual.
Augmentation potential	claude-haiku-4-5-20251001	4/5	AI assistants meaningfully augment faculty by handling routine recruitment outreach, filtering applications, and matching students to opportunities, freeing faculty time for personalized mentoring and strategic placement decisions.
Augmentation potential	claude-sonnet-5	3/5	AI tools can help draft recruitment materials, manage applicant data, and answer routine student inquiries, giving moderate productivity support to faculty involved in these activities.
Task automatability	claude-haiku-4-5-20251001	5/5	Student recruitment, registration, and placement involve largely structured, digital workflows (email outreach, form processing, data entry, matching students to programs) that current AI systems can automate end-to-end, achieving ≥50% time savings through automated communication, application screening, and placement matching.
Task automatability	claude-sonnet-5	2/5	This task involves relationship-building, advising, and institutional representation that requires human judgment and interpersonal presence; AI can support but not substitute for the bulk of the activity.'
Adoption barriers	claude-haiku-4-5-20251001	2/5	There are minimal legal or regulatory barriers to automating these administrative tasks; most are governed by institutional policy rather than licensing requirements, and student contact can be mediated or initiated by AI with minimal friction.
Adoption barriers	claude-sonnet-5	3/5	No strict licensing requirement, but institutional norms, personal rapport with prospective students, and internal governance over admissions decisions create moderate friction against automation.
Cost vs. human wage	claude-haiku-4-5-20251001	5/5	AI-driven recruitment, registration, and placement automation costs (chatbots, email campaigns, automated matching) are a small fraction of a full-time faculty member's loaded wage, making the economic case strongly favorable for substitution.
Cost vs. human wage	claude-sonnet-5	2/5	While automated messaging tools are cheap, the faculty involvement (interviews, advising, representing the department) still requires paid human time, so overall cost savings are limited.
Technical feasibility today	claude-haiku-4-5-20251001	4/5	Deployed products (CRM systems with AI routing, chatbots for recruitment queries, automated application processing, and job-matching algorithms) reliably handle these tasks in higher-education institutions today, though some judgmental edge cases (holistic admissions decisions) still benefit from human review.
Technical feasibility today	claude-sonnet-5	2/5	Some CRM and chatbot tools assist with recruitment communications and registration logistics, but no deployed product independently performs student recruitment or placement advising for faculty.

Compile, administer, and grade examinations, or assign this work to others.

CI 30–67 · exposure 50 · augmentation 88 · importance 4.4/5 · click for rater detail

Panel ratings per dimension for this task
Dimension	Model	Rating	Rationale
Sector adoption velocity	claude-haiku-4-5-20251001	2/5	Academic institutions are traditionally slow adopters of AI in assessment; while LMS auto-grading tools see some uptake, full displacement of faculty exam design and high-stakes grading remains rare in production, especially in STEM fields demanding nuanced judgment.
Sector adoption velocity	claude-sonnet-5	3/5	Higher education is adopting AI tools for grading and content generation at a moderate pace, with many pilots and increasing use of AI-assisted grading platforms, though widespread deep integration is still emerging.
Augmentation potential	claude-haiku-4-5-20251001	4/5	AI can meaningfully assist faculty by generating diverse question banks, auto-scoring multiple-choice sections, providing draft rubrics, and flagging outlier responses for review, substantially raising faculty productivity while the instructor retains final judgment on learning outcomes.
Augmentation potential	claude-sonnet-5	5/5	AI substantially helps instructors draft exam questions, build answer keys, and pre-grade assignments, significantly boosting productivity while the instructor retains final oversight.
Task automatability	claude-haiku-4-5-20251001	2/5	While AI can generate exam questions and auto-grade objective tests with reasonable accuracy, compiling coherent exams requiring domain expertise, administering them fairly (proctoring, accommodations, integrity monitoring), and grading complex physics problems with partial credit judgments requires human oversight. The full end-to-end workflow cannot meet the 50% time-saving bar without substantial human involvement.
Task automatability	claude-sonnet-5	4/5	AI can generate exam questions, create rubrics, and grade many response types (multiple choice, short answer, even essays with rubrics) with substantial time savings, though physics problem grading with partial credit for work shown still needs human oversight for edge cases.
Adoption barriers	claude-haiku-4-5-20251001	4/5	Postsecondary institutions have strong accreditation requirements, academic integrity obligations, and legal/contractual expectations that faculty assess student learning; delegating high-stakes exam grading to AI without faculty review faces institutional and liability friction.
Adoption barriers	claude-sonnet-5	2/5	Some institutional policies require instructor sign-off on final grades and academic integrity concerns exist, but no formal licensing barrier prevents AI-assisted grading or exam creation.
Cost vs. human wage	claude-haiku-4-5-20251001	2/5	Exam generation tools and basic auto-grading software are inexpensive, but comprehensive exam administration and quality review (especially for complex physics) still require faculty time, making all-in costs comparable to or exceeding the marginal cost of faculty oversight.
Cost vs. human wage	claude-sonnet-5	4/5	AI-based exam generation and grading assistance is far cheaper per unit of output than faculty or TA time, especially for objective or semi-structured problems.
Technical feasibility today	claude-haiku-4-5-20251001	3/5	Existing products (learning management systems, question banks, auto-grading tools) handle portions reliably in production, but integrated exam administration with fair proctoring and accurate grading of open-ended physics work remains error-prone and typically requires human validation and discretion.
Technical feasibility today	claude-sonnet-5	3/5	Products like AI-assisted grading tools and quiz generators (e.g., in LMS platforms, GPT-based tools) are used today, but grading complex physics derivations reliably at scale is not yet fully mature in production.

Evaluate and grade students' class work, laboratory work, assignments, and papers.

CI 25–47 · exposure 33 · augmentation 63 · importance 4.6/5 · click for rater detail

Panel ratings per dimension for this task
Dimension	Model	Rating	Rationale
Sector adoption velocity	claude-haiku-4-5-20251001	2/5	Higher education has low overall automation velocity; grading remains mostly manual despite decades of edtech. While some institutions pilot auto-grading systems, deep production adoption is limited compared to commercial or high-volume administrative sectors.
Sector adoption velocity	claude-sonnet-5	2/5	Higher education is a moderate-to-slow adopter of AI grading tools, with much experimentation but limited production-scale deployment for physics-specific assessment.
Augmentation potential	claude-haiku-4-5-20251001	3/5	AI can assist by generating draft rubrics, organizing submissions, flagging anomalies, and offering preliminary grading on objective items, which can speed faculty review workflows. However, the augmentation is mostly on the mechanics of grading rather than deepening pedagogical insight.
Augmentation potential	claude-sonnet-5	4/5	AI can efficiently pre-screen work, generate rubric-based feedback, flag errors, and draft comments, significantly speeding up an instructor's grading workflow while they retain final judgment.
Task automatability	claude-haiku-4-5-20251001	2/5	AI can partially assist with grading multiple-choice or short-answer physics problems with clear rubrics, but evaluating laboratory work, conceptual reasoning, and written explanations requires nuanced understanding of physics pedagogy and individual student progress that current systems struggle with reliably. End-to-end automation achieving 50% time savings at equal quality is not yet demonstrated at scale.
Task automatability	claude-sonnet-5	3/5	AI can grade objective/numeric physics problems and even provide feedback on written work, but nuanced grading of lab reports, proofs, and partial credit for physics reasoning still needs human judgment, so only part of the task meets the 50% time-savings bar reliably.
Adoption barriers	claude-haiku-4-5-20251001	4/5	Faculty have institutional and contractual autonomy over grading; universities are risk-averse about delegating grade assignment (which carries legal weight for transcripts and financial aid); and students expect human judgment and feedback on intellectual work. These create substantial friction against full substitution.
Adoption barriers	claude-sonnet-5	2/5	No licensing requirement blocks AI-assisted grading, though academic integrity policies, institutional accreditation standards, and instructor accountability for grades create moderate friction.
Cost vs. human wage	claude-haiku-4-5-20251001	2/5	AI grading platform fees plus integration labor and required human review oversight add up to meaningful cost; a faculty member's loaded salary for grading (~$30–50/hr for time spent) is often not yet undercut by the all-in cost of AI tools plus the fallback human auditing required.
Cost vs. human wage	claude-sonnet-5	3/5	AI grading assistance is cheap per assignment, but the need for instructor review of physics-specific reasoning and partial credit keeps effective cost comparable to human grading in many cases.
Technical feasibility today	claude-haiku-4-5-20251001	2/5	Products exist for auto-grading standardized assessments and homework platforms integrate AI scoring, but these are narrow in scope and require heavy manual setup of rubrics. Reliable grading of open-ended laboratory reports and papers—the cognitively demanding parts of the task—remains primarily manual in production education systems.
Technical feasibility today	claude-sonnet-5	2/5	Some LMS-integrated AI grading tools exist for STEM assignments, but they are not widely deployed for postsecondary physics grading involving lab work and open-ended derivations at scale with high reliability.

Advise students on academic and vocational curricula and on career issues.

CI 29–37 · exposure 25 · augmentation 75 · importance 3.7/5 · click for rater detail

Panel ratings per dimension for this task
Dimension	Model	Rating	Rationale
Sector adoption velocity	claude-haiku-4-5-20251001	2/5	Postsecondary institutions are slow to automate advising; most rely on human advisors or faculty, and pilot AI systems are rare and limited. The sector values face-to-face advising and is risk-averse about displacing personalized human guidance on critical career paths.
Sector adoption velocity	claude-sonnet-5	2/5	Higher education is a relatively slow-adopting sector for replacing human advising functions, though some administrative chat-based advising pilots exist.
Augmentation potential	claude-haiku-4-5-20251001	4/5	AI can meaningfully assist advisors by summarizing student records, retrieving curriculum and program information, suggesting relevant opportunities, and drafting responses—allowing human advisors to focus on judgment, relationship-building, and nuanced guidance. This is a strong augmentation use case.
Augmentation potential	claude-sonnet-5	4/5	AI can help faculty advisors quickly gather curriculum requirements, prep talking points, and generate career resource summaries, meaningfully boosting efficiency while the human retains the advising role.
Task automatability	claude-haiku-4-5-20251001	2/5	While AI can provide generic information about curricula and careers, advising students requires understanding their individual aptitudes, goals, and constraints, as well as delivering personalized guidance that accounts for nuance and interpersonal context. Current AI systems cannot reliably conduct the exploratory conversation and judgment that meaningful academic/career advising demands.
Task automatability	claude-sonnet-5	2/5	Advising involves relational, contextual judgment about individual students' goals, strengths, and institutional nuances that current AI cannot fully replicate end-to-end.dimensions.RepositoryOnly a portion (e.g., surfacing curriculum info or career resources) is automatable.
Adoption barriers	claude-haiku-4-5-20251001	4/5	Most institutions require that academic advising be delivered by credentialed advisors or faculty to ensure quality, accountability, and compliance with accreditation standards. Students and parents also strongly prefer human contact for consequential decisions, and institutional liability concerns limit full delegation to AI systems.
Adoption barriers	claude-sonnet-5	3/5	No strict licensing requirement to advise students, but institutional policies, liability concerns, and student/parent preference for human faculty advisors create moderate friction.
Cost vs. human wage	claude-haiku-4-5-20251001	4/5	The inference cost of an AI advisor is near-zero compared to the hourly rate of a faculty member or professional academic advisor. Even accounting for oversight and integration, the per-session cost differential is substantial.
Cost vs. human wage	claude-sonnet-5	3/5	AI can cheaply provide generic curriculum info, but for accurate, personalized advice requiring human oversight and correction, the cost advantage narrows considerably.
Technical feasibility today	claude-haiku-4-5-20251001	2/5	Chatbots exist that can deliver boilerplate career information and answer basic curriculum questions, but no product demonstrates the reliable judgment, understanding of individual student needs, and institutional knowledge required for postsecondary advising in production environments. Systems lack the depth to navigate complex trade-offs and student-specific constraints.
Technical feasibility today	claude-sonnet-5	2/5	Chatbot advising tools exist but are typically limited to FAQ-style guidance; nuanced academic/career advising for physics students in production is not demonstrated at scale.

Write grant proposals to procure external research funding.

CI 25–34 · exposure 25 · augmentation 75 · importance 3.7/5 · click for rater detail

Panel ratings per dimension for this task
Dimension	Model	Rating	Rationale
Sector adoption velocity	claude-haiku-4-5-20251001	2/5	Academic institutions and research offices are slow to adopt autonomous AI for grant writing; most current use is limited to drafting aids under close human supervision. Resistance stems from accountability norms, legal requirements, and concern that AI-generated proposals undermine research integrity and competitiveness.
Sector adoption velocity	claude-sonnet-5	2/5	Academic research administration and grant writing remain a slow-adopting niche within higher education, with individual faculty experimenting but institutional workflows and funder norms lagging behind faster-adopting sectors like tech or finance.
Augmentation potential	claude-haiku-4-5-20251001	4/5	AI writing tools significantly assist grant authors by drafting sections, improving prose, checking clarity, and accelerating iteration. Physics faculty regularly use ChatGPT and similar tools to draft background sections and refine language, substantially raising their writing productivity while they retain full control over research strategy and proposal content.
Augmentation potential	claude-sonnet-5	4/5	AI tools are already meaningfully useful for drafting boilerplate sections, editing prose, summarizing literature, and improving proposal clarity, significantly aiding the human author while they retain full ownership of scientific content and strategy.
Task automatability	claude-haiku-4-5-20251001	2/5	Grant writing involves highly customized arguments tailored to specific funding agency priorities, institutional context, and novel research claims. While AI can draft sections and improve language, the core requirement—persuasive articulation of original research vision and institutional fit—still requires substantial human judgment and domain expertise that current systems cannot reliably generate end-to-end at competitive quality.
Task automatability	claude-sonnet-5	2/5	AI can draft sections of a grant proposal but crafting a competitive, discipline-specific research narrative with novel scientific contributions, budget justification, and strategic framing still requires substantial human expertise and iteration.dim The task overall does not meet the 50% time-saving-at-equal-quality bar end-to-end.
Adoption barriers	claude-haiku-4-5-20251001	4/5	Grant proposals legally require institutional authorization (Grants & Contracts office approval, PI signature, institutional compliance with funder rules). Funding agencies expect human intellectual leadership and accountability; delegating proposal authorship to AI without human expertise creates liability and compliance risk, creating organizational and regulatory friction.
Adoption barriers	claude-sonnet-5	3/5	There's no legal requirement that a human write the proposal, but funding agencies expect PI-driven scientific vision and accountability, and reputational/institutional norms create moderate friction against wholesale AI authorship.
Cost vs. human wage	claude-haiku-4-5-20251001	2/5	AI tools are cheap per token, but integrating them into a grant-writing workflow requires oversight, human revision, and domain expert time. The loaded cost of a physicist's time spent reviewing, correcting, and strategizing AI drafts remains substantial relative to the marginal savings from partial automation.
Cost vs. human wage	claude-sonnet-5	3/5	AI drafting tools are cheap per use, but the overall cost of producing a fundable proposal still requires significant senior researcher time for technical content and strategy, making the net cost roughly comparable to unaided human effort with modest savings.
Technical feasibility today	claude-haiku-4-5-20251001	2/5	AI writing assistants (ChatGPT, Claude) exist and can help with drafts and editing, but no deployed product reliably produces submission-ready grant proposals independently. Research offices still require significant human-led revision, fact-checking, and institutional customization before submission, and competitive success depends on human strategy and insight.
Technical feasibility today	claude-sonnet-5	2/5	Products like ChatGPT and specialized writing assistants are used informally by academics to draft proposal text, but no mature deployed product reliably produces fundable physics grant proposals without heavy expert revision.

Select and obtain materials and supplies, such as textbooks and laboratory equipment.

CI 23–35 · exposure 25 · augmentation 50 · importance 3.5/5 · click for rater detail

Panel ratings per dimension for this task
Dimension	Model	Rating	Rationale
Sector adoption velocity	claude-haiku-4-5-20251001	1/5	Higher education remains largely traditional in procurement automation, with human department staff managing material selection and ordering as a standard practice. Adoption of AI-driven procurement in academic settings is minimal compared to corporate environments.
Sector adoption velocity	claude-sonnet-5	2/5	Higher education administrative and instructional support functions show slow AI adoption, and procurement of physical lab equipment is not a digitized workflow prioritized in current AI rollouts.
Augmentation potential	claude-haiku-4-5-20251001	3/5	AI could usefully assist by researching vendors, comparing specifications, summarizing product reviews, and drafting procurement requisitions, helping humans make faster decisions on material selection without fully automating the approval and ordering process.
Augmentation potential	claude-sonnet-5	3/5	AI tools can help compare textbook options, generate supply lists, and search vendor catalogs, providing moderate assistance to the teacher who still finalizes and orders materials.
Task automatability	claude-haiku-4-5-20251001	2/5	While AI could assist in researching, comparing, and drafting procurement requests for materials, the task requires judgment about specific departmental needs, budget constraints, and supplier relationships that demand human oversight. End-to-end automation would require navigating procurement systems, vendor approval, and inventory management—feasible only with significant integration work.
Task automatability	claude-sonnet-5	2/5	AI can help identify and compare textbooks or suggest lab equipment lists, but final selection requires judgment about course fit, budget, and physical procurement/logistics that AI cannot execute end-to-end.paksowa Ordering and physically obtaining supplies still requires human action.
Adoption barriers	claude-haiku-4-5-20251001	4/5	Educational institutions have formal procurement policies, budget authorization requirements, and approval workflows that create legal and administrative barriers. Ordering equipment often requires human sign-off from department heads or procurement officers, and vendor contracts typically require authorized human parties.
Adoption barriers	claude-sonnet-5	2/5	No licensing barrier, but institutional purchasing processes, budget approval, and departmental preferences create moderate organizational friction against full automation.
Cost vs. human wage	claude-haiku-4-5-20251001	2/5	The cost of an AI system with integration to procurement platforms and ongoing oversight would likely exceed the time savings from automating a task that a department administrator might handle as one part of their broader responsibilities, especially in academic contexts.
Cost vs. human wage	claude-sonnet-5	2/5	AI assistance for research/comparison is cheap, but the overall task still requires human decision-making, vendor interaction, and physical procurement, keeping overall costs comparable to a human doing the task.
Technical feasibility today	claude-haiku-4-5-20251001	2/5	No deployed products reliably handle complete procurement workflows for educational institutions end-to-end. Procurement systems exist but typically require manual review, approval chains, and human decision-making on equipment selection criteria specific to pedagogical goals.
Technical feasibility today	claude-sonnet-5	2/5	No deployed product autonomously selects and procures physics course materials and lab equipment; existing tools (search engines, recommendation systems) only assist with research, not execution.

Plan, evaluate, and revise curricula, course content, course materials, and methods of instruction.

CI 23–34 · exposure 25 · augmentation 63 · importance 4.1/5 · click for rater detail

Panel ratings per dimension for this task
Dimension	Model	Rating	Rationale
Sector adoption velocity	claude-haiku-4-5-20251001	1/5	Higher education adoption of AI for core curriculum development remains minimal in production settings. Universities are conservative stewards of their curricula, and faculty resistance to algorithmic curriculum design is substantial despite increasing experimentation with AI tools.
Sector adoption velocity	claude-sonnet-5	2/5	Higher education adopts AI tools unevenly and cautiously, with pilots for content generation but slow institutional change in official curriculum processes.
Augmentation potential	claude-haiku-4-5-20251001	3/5	AI can usefully assist faculty by generating content suggestions, organizing materials, and proposing course structures that faculty then refine and adapt, thereby accelerating the planning process while the expert remains in control of pedagogical decisions.
Augmentation potential	claude-sonnet-5	4/5	AI tools substantially assist in drafting course materials, generating problem sets, and suggesting content updates, meaningfully speeding up the work while the instructor retains final judgment.
Task automatability	claude-haiku-4-5-20251001	2/5	AI can draft course outlines and suggest learning materials, but cannot independently evaluate learning outcomes, adapt content to student cohorts, or make pedagogical judgments about instructional methods without substantial human oversight. The evaluative and revisions components require disciplinary expertise and understanding of student needs.
Task automatability	claude-sonnet-5	2/5	AI can draft syllabi and suggest content, but planning and revising curricula requires institutional judgment, accreditation alignment, and pedagogical expertise that current AI cannot fully replace end-to-end.atura
Adoption barriers	claude-haiku-4-5-20251001	4/5	Faculty curriculum authority is deeply entrenched in academic governance, accreditation standards, and professional norms. Institutions typically require faculty approval of curriculum changes, and academic freedom protections make automated substitution both legally and culturally infeasible.
Adoption barriers	claude-sonnet-5	3/5	No formal licensing requirement for curriculum design specifically, but institutional accreditation processes, faculty governance, and departmental review create real friction against full automation.
Cost vs. human wage	claude-haiku-4-5-20251001	2/5	The cost of AI tools plus required expert human review and revision approaches the loaded cost of faculty time spent on curriculum work, especially given the need for domain expertise validation and institutional knowledge integration.
Cost vs. human wage	claude-sonnet-5	3/5	AI can cheaply generate drafts of materials, lowering some costs, but human oversight, subject expertise, and institutional review remain necessary, keeping overall cost comparable to human-led effort.
Technical feasibility today	claude-haiku-4-5-20251001	2/5	While AI tools exist for generating course content and lesson plans, no deployed product reliably performs the full end-to-end task of curriculum planning, evaluation, and revision with equal quality to expert faculty judgment. Most systems require significant human direction and validation.
Technical feasibility today	claude-sonnet-5	2/5	Products like ChatGPT or course-design assistants exist and are used for drafting materials, but no deployed product autonomously plans and revises entire postsecondary physics curricula in production.

Keep abreast of developments in the field by reading current literature, talking with colleagues, and participating in professional conferences.

CI 16–38 · exposure 17 · augmentation 75 · importance 4.0/5 · click for rater detail

Panel ratings per dimension for this task
Dimension	Model	Rating	Rationale
Sector adoption velocity	claude-haiku-4-5-20251001	2/5	While academia has adopted literature-tracking and visualization tools, the core task of staying current through active reading and conference presence remains stubbornly human-centered. Adoption of AI agents to replace this responsibility remains minimal; the sector views it as a core faculty responsibility.
Sector adoption velocity	claude-sonnet-5	3/5	Higher education and academic research show moderate AI tool adoption (literature review assistants, research databases with AI search) but full replacement of professional engagement is rare.
Augmentation potential	claude-haiku-4-5-20251001	4/5	AI powerfully augments this task: automated literature alerts, paper summarization, conference abstract filtering, and recommendation systems substantially enhance a physicist's ability to scan the field efficiently and identify relevant work for deeper reading. The human remains the curator and synthesizer but becomes much more productive.
Augmentation potential	claude-sonnet-5	4/5	AI literature summarization, alerting services, and research assistants significantly help academics track new developments faster than manual reading alone.
Task automatability	claude-haiku-4-5-20251001	1/5	Staying current through reading, collegial dialogue, and conference participation requires sustained professional judgment, contextual synthesis, and authentic engagement. Current AI cannot replicate the selective filtering and integrative reasoning a physicist applies when evaluating which developments matter for their subfield and teaching.
Task automatability	claude-sonnet-5	2/5	AI can help surface and summarize literature, but the actual task of staying current requires ongoing human judgment, networking, and synthesis that isn't fully replaceable end-to-end.
Adoption barriers	claude-haiku-4-5-20251001	4/5	Professional norms, institutional culture, and career advancement expectations strongly bind this task to human engagement. Tenure, promotion, and credibility in a discipline depend visibly on the professor's own intellectual leadership and collegial presence—substituting AI would damage both institutional standing and professional identity.
Adoption barriers	claude-sonnet-5	2/5	No licensing barrier prevents AI assistance, but professional norms and value of human networking/collegial exchange create moderate friction against full substitution.
Cost vs. human wage	claude-haiku-4-5-20251001	2/5	AI tools (literature monitoring, summarization) have low per-unit cost, but the human time investment in reading, synthesis, and conference attendance remains substantial and non-negotiable. The all-in cost of maintaining a professor's professional development far exceeds the marginal cost of automated alerts.
Cost vs. human wage	claude-sonnet-5	2/5	AI tools for literature scanning are cheap, but they don't replace the full task including networking and conference attendance, so cost comparison for the whole task favors humans still doing most of the work.
Technical feasibility today	claude-haiku-4-5-20251001	2/5	While AI can summarize papers and identify new research, no deployed system can reliably perform the full task of staying abreast—which requires curating what matters, engaging in substantive colleague conversations, and gaining value from live conference interaction. Deployed tools assist components but do not execute the integrated professional awareness task.
Technical feasibility today	claude-sonnet-5	2/5	AI literature summarization and alerting tools exist and are used by researchers, but no deployed product autonomously 'keeps abreast' of a field including conference participation and colleague discussion.

Prepare and deliver lectures to undergraduate or graduate students on topics such as quantum mechanics, particle physics, and optics.

CI 21–25 · exposure 25 · augmentation 75 · importance 4.4/5 · click for rater detail

Panel ratings per dimension for this task
Dimension	Model	Rating	Rationale
Sector adoption velocity	claude-haiku-4-5-20251001	2/5	Adoption remains minimal in postsecondary education despite digitization elsewhere; institutions have moved toward hybrid and online delivery, but these typically still require faculty-led instruction rather than AI replacement; pilots exist but production deployment of AI as primary lecturer is rare and often limited to supplementary content.
Sector adoption velocity	claude-sonnet-5	2/5	Higher education adopts AI tools for tutoring and content support but has been slow to replace core lecture delivery, given tenure structures, accreditation, and pedagogical norms.
Augmentation potential	claude-haiku-4-5-20251001	3/5	AI can meaningfully assist faculty by drafting lecture notes, generating problem sets, creating visualizations, and providing real-time student question answering during live sessions, raising overall preparation and delivery efficiency while keeping the instructor in the loop.
Augmentation potential	claude-sonnet-5	5/5	AI is highly useful for drafting lecture notes, generating problem sets, creating visualizations, and explaining complex quantum/particle physics concepts, significantly aiding lecture preparation.
Task automatability	claude-haiku-4-5-20251001	2/5	While AI can generate lecture outlines, notes, and visual aids, delivering a live lecture requires real-time interaction, responsiveness to student confusion, pedagogical judgment, and the ability to adapt content on the fly—capabilities current systems cannot reliably handle end-to-end. Automated lecture generation might save some preparation time but falls well short of the 50% threshold for full task automation.
Task automatability	claude-sonnet-5	2/5	AI can help draft lecture content and explanations, but live delivery, adapting to student questions, and maintaining classroom engagement require human presence and judgment not replaceable end-to-end today.
Adoption barriers	claude-haiku-4-5-20251001	4/5	Strong barriers exist: accreditation bodies, institutional policies, and student expectations generally require faculty presence and accountability for instruction; institutions have reputational and legal liability for educational quality; and many jurisdictions have specific faculty credential requirements for course delivery and grading authority.
Adoption barriers	claude-sonnet-5	4/5	Accreditation, degree-granting requirements, and institutional norms generally require a qualified instructor of record to teach and be accountable for course content, creating strong organizational and regulatory friction.
Cost vs. human wage	claude-haiku-4-5-20251001	1/5	Even accounting for AI content generation and lecture support, the per-student cost of AI systems plus required faculty oversight typically exceeds the cost of direct faculty instruction, especially at the postsecondary level where lectures are already high-value, knowledge-intensive events.
Cost vs. human wage	claude-sonnet-5	2/5	While content generation is cheap, actual lecture delivery still requires a human instructor or costly video/AI-avatar production pipeline, so all-in cost is not clearly cheaper than an existing faculty member's marginal lecture time.
Technical feasibility today	claude-haiku-4-5-20251001	2/5	No deployed product reliably delivers complete lectures to students with the depth, real-time interactivity, and educational judgment required. AI can assist with content preparation and answer simple student questions, but production systems do not yet perform the full task—live instruction with adaptive pedagogy—at acceptable quality in real academic settings.
Technical feasibility today	claude-sonnet-5	2/5	Tools like ChatGPT or Khan Academy-style AI tutors can generate explanations of physics topics, but no deployed product reliably delivers full undergraduate/graduate physics lectures in place of an instructor.

Conduct research in a particular field of knowledge and publish findings in professional journals, books, or electronic media.

CI 0–32 · exposure 13 · augmentation 75 · importance 3.7/5 · click for rater detail

Panel ratings per dimension for this task
Dimension	Model	Rating	Rationale
Sector adoption velocity	claude-haiku-4-5-20251001	1/5	Research publication in academia remains a fundamentally human-driven, slow-moving sector with strong norms around originality and human authorship. Adoption of AI for autonomous research and publication is effectively nonexistent in production environments.
Sector adoption velocity	claude-sonnet-5	3/5	Academic research increasingly uses AI tools for literature synthesis, coding, and manuscript drafting, but full-scale production adoption for autonomous research generation remains uneven and largely experimental.
Augmentation potential	claude-haiku-4-5-20251001	4/5	AI substantially assists postsecondary researchers through literature synthesis, data analysis, manuscript drafting and editing, and visualization—significantly accelerating productivity on research tasks while the human researcher retains intellectual leadership and validation responsibility.
Augmentation potential	claude-sonnet-5	4/5	AI substantially accelerates literature reviews, data analysis, code writing, and drafting of papers, meaningfully boosting researcher productivity while the physicist retains control over hypotheses, experiments, and interpretation.
Task automatability	claude-haiku-4-5-20251001	1/5	Publishing original research findings requires novel intellectual contribution, domain expertise, creative hypothesis formation, and peer-review navigation—capabilities that current AI systems cannot reliably perform end-to-end. While AI can assist with writing drafts or literature review, the core research conception and validation remain fundamentally human intellectual work.
Task automatability	claude-sonnet-5	2/5	Original physics research requires designing experiments, running novel theoretical derivations, and generating genuinely new knowledge—AI can assist with literature review, code, and drafting but cannot independently conduct research or ensure valid, publishable findings end-to-end.
Adoption barriers	claude-haiku-4-5-20251001	5/5	Research integrity, academic freedom, and authorship carry legal, institutional, and professional liability. Publication requires human accountability; journals demand author sign-off and institutions hold researchers legally and ethically responsible for findings. No automation can substitute for this human-certified contribution.
Adoption barriers	claude-sonnet-5	3/5	No strict licensing barrier prevents AI-assisted research, but norms around authorship, peer review, research integrity, and academic credentialing create meaningful friction against full automation.
Cost vs. human wage	claude-haiku-4-5-20251001	1/5	The cost of conducting original research (including equipment, data collection, and human expertise) vastly exceeds what AI inference could contribute; moreover, AI cannot yet replace the human researcher's intellectual labor that commands a professor's salary.
Cost vs. human wage	claude-sonnet-5	2/5	While AI tools cut costs for certain subtasks (searching, drafting, coding), the overall research process still requires substantial expert human time for experimentation, validation, and peer-reviewed writing, keeping all-in AI substitution costs comparable to or higher than expected savings.
Technical feasibility today	claude-haiku-4-5-20251001	1/5	No deployed product conducts original research and publishes findings independently. AI tools exist for writing support and data analysis, but no production system autonomously performs the full research cycle from problem formulation through peer-reviewed publication.
Technical feasibility today	claude-sonnet-5	2/5	Deployed tools (literature search assistants, code generation, writing aids) support pieces of the research pipeline, but no product reliably conducts full original physics research and produces publishable findings unsupervised.

Initiate, facilitate, and moderate classroom discussions.

CI 0–20 · exposure 8 · augmentation 38 · importance 4.1/5 · click for rater detail

Panel ratings per dimension for this task
Dimension	Model	Rating	Rationale
Sector adoption velocity	claude-haiku-4-5-20251001	1/5	Postsecondary institutions remain highly resistant to automating core classroom instruction; discussion facilitation is seen as central to the teaching role. Adoption of AI in these settings is slow, pilot-stage, and limited to supplementary tasks, not live classroom moderation.
Sector adoption velocity	claude-sonnet-5	2/5	Higher education is adopting AI for content creation and grading assistance but adoption of AI to replace live discussion facilitation remains rare and mostly experimental.
Augmentation potential	claude-haiku-4-5-20251001	2/5	AI could assist a teacher by generating discussion prompts, summarizing key points, or identifying common student misconceptions post-session, but it offers minimal real-time assistance during the actual moderation of discussion where judgment and presence are most critical.
Augmentation potential	claude-sonnet-5	3/5	AI can help generate discussion prompts, summarize readings, or provide real-time content lookups, offering moderate assistance to the instructor without replacing the live facilitation role.
Task automatability	claude-haiku-4-5-20251001	1/5	Classroom discussion moderation requires real-time social perception, judgment about student engagement and emotional states, dynamic redirection of conversation, and building rapport—capabilities that current AI systems cannot perform reliably in live classroom settings. No off-the-shelf system can replace a human moderator and achieve 50% time savings at equal educational quality.
Task automatability	claude-sonnet-5	2/5	Leading a live, adaptive classroom discussion requires reading social dynamics, real-time judgment, and rapport-building that current AI cannot replicate end-to-end in a physical classroom setting.6ai can propose discussion questions but cannot run the interactive session itself with equal quality.6ai tools do not substitute for the live facilitation role.
Adoption barriers	claude-haiku-4-5-20251001	5/5	There is a hard institutional and legal requirement that postsecondary instruction be conducted by qualified, credentialed instructors who are accountable for student learning and classroom dynamics. Teachers cannot legally delegate core pedagogical functions like discussion facilitation to an automated system without human oversight.
Adoption barriers	claude-sonnet-5	4/5	Accreditation, instructor-of-record requirements, and the expectation of a live human educator create strong institutional and pedagogical barriers to replacing this interactive teaching function.
Cost vs. human wage	claude-haiku-4-5-20251001	2/5	Even if a system could moderate discussions (which it cannot reliably), integrating it into a live classroom would require significant infrastructure, oversight, and coordination costs that would approach or exceed the salary cost of a teaching assistant or instructor managing the discussion.
Cost vs. human wage	claude-sonnet-5	1/5	There is no viable AI substitute performing this exact live facilitation function at scale, so cost comparison favors the human since AI cannot yet deliver an equivalent output.
Technical feasibility today	claude-haiku-4-5-20251001	1/5	While AI chatbots can simulate discussion, no deployed product reliably moderates live classroom discussions with the contextual awareness, conflict resolution, and pedagogical responsiveness that the task demands. Existing systems are research-stage or narrow classroom simulations, not production classroom moderators.
Technical feasibility today	claude-sonnet-5	1/5	No deployed product autonomously initiates and moderates live in-person postsecondary classroom discussions; existing AI tools (chatbots, discussion-board moderation) exist for online asynchronous contexts but not for live faculty-led seminar facilitation.

Maintain and repair laboratory equipment.

CI 5–15 · exposure 5 · augmentation 38 · importance 3.4/5 · click for rater detail

Panel ratings per dimension for this task
Dimension	Model	Rating	Rationale
Sector adoption velocity	claude-haiku-4-5-20251001	1/5	Academic institutions operate with legacy equipment and established technician workforces; adoption of automation for physical repair is minimal, and the sector is slow to digitize maintenance operations beyond scheduling and inventory systems.
Sector adoption velocity	claude-sonnet-5	1/5	Physical maintenance work in academic labs is a low-digitization, hands-on domain with essentially no AI/robotic adoption for equipment repair currently.
Augmentation potential	claude-haiku-4-5-20251001	3/5	AI can meaningfully assist technicians by retrieving equipment manuals, suggesting diagnostic steps, and organizing maintenance records, thereby reducing time spent searching for documentation and improving repair efficiency without replacing human expertise.
Augmentation potential	claude-sonnet-5	2/5	AI can help by providing troubleshooting guidance, manuals, or diagnostic checklists, but it does not meaningfully transform the physical repair process itself.
Task automatability	claude-haiku-4-5-20251001	1/5	Maintaining and repairing laboratory equipment requires hands-on diagnostics, physical manipulation, and real-time problem-solving in varied equipment contexts. Current AI systems cannot physically disassemble, test, or repair equipment, nor can they reliably diagnose complex mechanical/electrical faults without human inspection.
Task automatability	claude-sonnet-5	1/5	Physical diagnosis, calibration, and repair of laboratory instruments requires hands-on manipulation and troubleshooting that current AI systems cannot perform without robotic embodiment, which is not generally available.
Adoption barriers	claude-haiku-4-5-20251001	4/5	Safety regulations, equipment warranties, institutional liability for improper repairs, and the requirement for qualified technicians to certify equipment functionality create strong legal and organizational barriers to automation of repair tasks.
Adoption barriers	claude-sonnet-5	3/5	No licensing requirement specifically bars AI, but physical safety concerns (electrical, optical, chemical equipment) and institutional liability create moderate friction against unsupervised automation.
Cost vs. human wage	claude-haiku-4-5-20251001	1/5	AI cannot perform the core physical repair work, so the cost comparison is moot; AI can only marginally reduce the technician's time through information retrieval, which does not justify a favorable ratio compared to skilled labor doing the actual repair.
Cost vs. human wage	claude-sonnet-5	1/5	There is no viable AI substitute for physical repair work, so the human remains the only cost-effective option for this task.
Technical feasibility today	claude-haiku-4-5-20251001	2/5	While AI can assist with documentation lookup and troubleshooting guides via language models, no deployed products can independently diagnose and repair laboratory equipment. AI might support technicians by retrieving manuals or suggesting checks, but actual maintenance and repair remain human-dependent.
Technical feasibility today	claude-sonnet-5	1/5	No deployed product autonomously maintains or repairs physics lab equipment; this remains firmly a manual, hands-on task performed by humans.

Collaborate with colleagues to address teaching and research issues.

CI 5–13 · exposure 0 · augmentation 38 · importance 4.0/5 · click for rater detail

Panel ratings per dimension for this task
Dimension	Model	Rating	Rationale
Sector adoption velocity	claude-haiku-4-5-20251001	1/5	Higher education remains a laggard sector for process automation. Collaboration on teaching and research is core to academic identity and institutional function; adoption of AI substitution is minimal and not accelerating.
Sector adoption velocity	claude-sonnet-5	2/5	Higher education adopts AI slowly for interpersonal academic governance and collaboration, even though it may adopt AI tools for other administrative tasks faster.
Augmentation potential	claude-haiku-4-5-20251001	2/5	AI can assist marginally—summarizing research, drafting discussion frameworks, or organizing meeting notes—but cannot replace the substantive intellectual exchange, trust-building, and shared expertise that colleagues bring to addressing teaching and research challenges together.
Augmentation potential	claude-sonnet-5	3/5	AI can help draft agendas, summarize meeting notes, or organize shared research materials, providing moderate assistance around the edges of collaboration.
Task automatability	claude-haiku-4-5-20251001	1/5	This task requires human-to-human interpersonal collaboration, consensus-building, and contextual judgment about institutional teaching and research priorities. Current AI cannot independently establish trust, negotiate conflicts, or make substantive decisions in peer collaboration without human leadership.
Task automatability	claude-sonnet-5	1/5	This is inherently interpersonal collegial collaboration requiring shared institutional context, trust-building, and negotiation that AI cannot substitute for end-to-end.'
Adoption barriers	claude-haiku-4-5-20251001	4/5	Academic governance, shared intellectual work, and pedagogical decisions are deeply embedded in collegial norms and institutional structures. Organizational culture strongly favors human collaboration; effective teaching and research advancement depend on buy-in from human colleagues, creating substantial friction against full automation.
Adoption barriers	claude-sonnet-5	3/5	No formal licensing requirement, but strong organizational and social norms around faculty governance and peer collaboration create real friction against substitution.
Cost vs. human wage	claude-haiku-4-5-20251001	1/5	The task is fundamentally about human professional interaction and mutual problem-solving. AI tools could assist at margins (note-taking, documentation), but cannot replace the human effort; all-in cost per outcome remains higher than human execution.
Cost vs. human wage	claude-sonnet-5	1/5	There is no AI product performing this task, so no meaningful cost comparison exists; human collaboration remains the only viable option.
Technical feasibility today	claude-haiku-4-5-20251001	1/5	No deployed product replaces colleague-to-colleague collaboration on teaching and research issues. AI might summarize discussions or draft agendas, but it does not perform the core interpersonal and decision-making work that defines this task.
Technical feasibility today	claude-sonnet-5	1/5	No deployed product performs collegial collaboration on teaching/research issues; AI tools at best support communication logistics, not the substantive collaboration itself.

Maintain regularly scheduled office hours to advise and assist students.

CI 0–16 · exposure 0 · augmentation 50 · importance 4.2/5 · click for rater detail

Panel ratings per dimension for this task
Dimension	Model	Rating	Rationale
Sector adoption velocity	claude-haiku-4-5-20251001	1/5	Postsecondary institutions have shown minimal adoption of AI to replace office hours; higher education is a laggard sector in automation of human-facing advising, with strong organizational and cultural resistance.
Sector adoption velocity	claude-sonnet-5	2/5	Higher education adopts AI tools slowly for direct student-facing advising roles, though usage of chatbots for administrative FAQs is growing modestly.
Augmentation potential	claude-haiku-4-5-20251001	3/5	AI tools (tutoring bots, scheduling assistants, prior-question databases) can assist professors in managing office hour logistics and answering routine questions, but the core task remains faculty-driven interpersonal advising.
Augmentation potential	claude-sonnet-5	3/5	AI tools can help prep materials, draft answers to common questions, or provide supplementary tutoring resources students access outside office hours, moderately aiding the overall advising function.
Task automatability	claude-haiku-4-5-20251001	1/5	Office hours require real-time, contextual human engagement with students—answering individual questions, gauging understanding, and providing personalized academic mentorship. Current AI cannot reliably replicate the adaptive, empathetic judgment needed, and no system offers ≥50% time savings while maintaining equal quality.
Task automatability	claude-sonnet-5	1/5	Office hours require live, personalized human presence, mentorship, and relationship-building that current AI cannot replicate as an end-to-end substitute for the scheduled interaction itself.
Adoption barriers	claude-haiku-4-5-20251001	5/5	Academic advising and direct student support are inherently human-contact functions. Institutions and accreditors expect faculty to maintain office hours as a regulatory and contractual expectation; liability and student outcomes are tied to human judgment and presence.
Adoption barriers	claude-sonnet-5	3/5	No licensing barrier prevents AI-assisted advising, but institutional norms, accreditation expectations, and student preference for human mentorship create meaningful friction.
Cost vs. human wage	claude-haiku-4-5-20251001	1/5	Deploying a working AI system to replace office hours (including oversight, error correction, and student relationship management) would cost significantly more than the instructor's marginal time cost for holding regular hours.
Cost vs. human wage	claude-sonnet-5	2/5	While an AI chatbot could handle simple queries cheaply, the actual task of holding office hours has no comparable AI equivalent, so cost comparison mostly doesn't apply to full substitution.
Technical feasibility today	claude-haiku-4-5-20251001	1/5	No deployed product reliably conducts office hours as a human professor would. Chatbots can answer generic questions but cannot provide the individualized advising, problem-solving clarification, and relational continuity that defines this task.
Technical feasibility today	claude-sonnet-5	1/5	No deployed product actually holds office hours as a substitute for faculty; chatbots exist for FAQ-style support but not as a replacement for scheduled advising sessions.

Provide professional consulting services to government or industry.

CI 4–11 · exposure 0 · augmentation 63 · importance 2.3/5 · click for rater detail

Panel ratings per dimension for this task
Dimension	Model	Rating	Rationale
Sector adoption velocity	claude-haiku-4-5-20251001	1/5	Government and industrial physics consulting remains a high-trust, relationship-driven domain with strong preference for human experts bearing professional responsibility. Adoption of autonomous AI consulting in these sectors is minimal; organizations continue to hire and rely on human consultants.
Sector adoption velocity	claude-sonnet-5	2/5	Academic consulting is a small, bespoke, relationship-driven market with limited AI agent deployment compared to fast-adopting sectors like finance or software.
Augmentation potential	claude-haiku-4-5-20251001	3/5	AI can assist consultants by accelerating literature reviews, modeling scenarios, organizing data, and drafting technical sections of reports. However, the consultant must integrate, validate, and take professional responsibility for all recommendations, limiting AI's autonomous contribution to supporting research and drafting phases.
Augmentation potential	claude-sonnet-5	4/5	AI tools can meaningfully assist with literature review, data analysis, report drafting, and modeling, boosting a consultant's productivity while the expert retains final judgment.
Task automatability	claude-haiku-4-5-20251001	1/5	Consulting requires deep domain expertise, judgment, client relationship management, and customized problem-solving that cannot be reduced to a routine algorithmic process. Current AI systems lack the credibility, accountability, and contextual sophistication to serve as primary consultants in physics-related government or industry matters.
Task automatability	claude-sonnet-5	1/5	Consulting requires deep domain expertise, contextual judgment, relationship-building, and accountability that current AI cannot deliver end-to-end for a client engagement.
Adoption barriers	claude-haiku-4-5-20251001	5/5	Professional consulting, especially to government agencies, typically requires licensed experts, contractual accountability, liability insurance, and client certification of consultant qualifications. Legal and regulatory frameworks mandate human expert sign-off and assume human judgment; automation cannot legally replace this.
Adoption barriers	claude-sonnet-5	4/5	Government and industry consulting typically requires credentialed expertise, contractual liability, and named accountability, creating strong barriers to full AI substitution.
Cost vs. human wage	claude-haiku-4-5-20251001	2/5	AI inference is cheap, but the full cost structure of consulting includes domain validation, liability coverage, integration overhead, and human oversight to ensure recommendations meet professional standards. These overheads narrow the economic advantage, though AI-assisted research may reduce some billable hours.
Cost vs. human wage	claude-sonnet-5	2/5	AI can cut some research and drafting time, but the bulk of value (expert judgment, credibility, liability, client trust) still requires a paid human expert, keeping overall cost comparable.
Technical feasibility today	claude-haiku-4-5-20251001	1/5	No deployed AI product reliably functions as an independent professional consultant to government or industry in physics. While AI can assist with research or technical generation, autonomous consulting—which requires legal responsibility, client trust, and bespoke strategic guidance—remains outside production capabilities.
Technical feasibility today	claude-sonnet-5	1/5	No deployed product independently provides professional physics consulting services to government or industry; AI is at most a research/drafting aid used by the human consultant.

Supervise undergraduate or graduate teaching, internship, and research work.

CI 0–13 · exposure 5 · augmentation 50 · importance 4.2/5 · click for rater detail

Panel ratings per dimension for this task
Dimension	Model	Rating	Rationale
Sector adoption velocity	claude-haiku-4-5-20251001	1/5	Academic institutions have shown minimal adoption of AI for core supervision responsibilities; traditional models of faculty mentorship and oversight remain deeply embedded in postsecondary pedagogy and governance structures.
Sector adoption velocity	claude-sonnet-5	2/5	Higher education adopts AI tools slowly for core mentorship and supervisory functions, though usage of AI for administrative support is growing modestly.
Augmentation potential	claude-haiku-4-5-20251001	3/5	AI can usefully assist with administrative scheduling, progress tracking, and preliminary draft feedback on student work, moderately increasing supervisor efficiency while the faculty member retains full supervisory responsibility and judgment.
Augmentation potential	claude-sonnet-5	3/5	AI can help physics faculty by drafting feedback, tracking research progress, suggesting resources, or analyzing student work, meaningfully aiding but not replacing the supervisory relationship.
Task automatability	claude-haiku-4-5-20251001	1/5	Supervising teaching, internships, and research work requires real-time judgment of student progress, motivation, and safety—particularly in laboratory settings—along with adaptive coaching and mentoring that current AI systems cannot reliably replicate end-to-end.
Task automatability	claude-sonnet-5	1/5	Supervising students' hands-on teaching, internships, and research requires ongoing personal mentorship, judgment calls about lab safety, and relationship-based feedback that current AI cannot perform end-to-end.
Adoption barriers	claude-haiku-4-5-20251001	5/5	Institutional policy, accreditation standards, and duty-of-care liability require a qualified human faculty member to directly supervise students, especially in research and laboratory contexts; legal and regulatory frameworks mandate personal human oversight.
Adoption barriers	claude-sonnet-5	4/5	Institutional accreditation, faculty employment requirements, and academic mentorship norms require a qualified human faculty member to supervise students, creating strong structural barriers.
Cost vs. human wage	claude-haiku-4-5-20251001	1/5	The integration and oversight costs of AI systems that merely assist with administrative tasks would exceed the value delivered, and would still require a human supervisor to perform the core supervision function.
Cost vs. human wage	claude-sonnet-5	1/5	There is no viable AI substitute performing this supervisory role, so cost comparison favors the human by default since AI cannot deliver the output.
Technical feasibility today	claude-haiku-4-5-20251001	2/5	While AI can assist with scheduling, grading support, and documentation, no deployed system can autonomously supervise the academic and safety dimensions of student work; existing products lack the situated judgment needed for meaningful supervision.
Technical feasibility today	claude-sonnet-5	1/5	No deployed product supervises student research or teaching internships autonomously; at best AI tools assist with scheduling or feedback drafting, not supervision itself.

Perform administrative duties, such as serving as department head.

CI 0–11 · exposure 8 · augmentation 50 · importance 3.2/5 · click for rater detail

Panel ratings per dimension for this task
Dimension	Model	Rating	Rationale
Sector adoption velocity	claude-haiku-4-5-20251001	1/5	Higher education institutions have shown minimal appetite for automating department administration; these roles remain firmly human-controlled and filled by faculty. The sector's conservative governance structures and the legal accountability embedded in the role create structural resistance to automation.
Sector adoption velocity	claude-sonnet-5	2/5	Higher education administration adopts AI slowly for governance and leadership functions, though it uses AI tools for peripheral tasks like scheduling or reporting.
Augmentation potential	claude-haiku-4-5-20251001	3/5	AI tools can meaningfully assist department heads with meeting scheduling, expense tracking, document drafting, and data analysis on student or faculty metrics, improving administrative efficiency. However, the irreducible human components (final decisions, relationships, accountability) limit the transformative impact.
Augmentation potential	claude-sonnet-5	3/5	AI can assist with scheduling, drafting reports, budget analysis, and email management that support the administrative workload, though the core leadership responsibilities remain human-driven.
Task automatability	claude-haiku-4-5-20251001	2/5	Administrative duties for department heads involve complex interpersonal negotiation, budget allocation decisions, hiring judgment, and institutional accountability that current AI cannot fully automate. While AI could assist with scheduling, record-keeping, and routine correspondence, the core responsibilities (personnel decisions, strategic planning, conflict resolution) require sustained human judgment and authority.
Task automatability	claude-sonnet-5	1/5	Serving as department head involves leadership, personnel decisions, budget authority, and institutional politics that require human judgment, relationships, and accountability that current AI cannot execute end-to-end.
Adoption barriers	claude-haiku-4-5-20251001	5/5	Department head roles are legally and contractually bound to specific individuals who must hold institutional authority, sign official documents, and bear responsibility for budget, hiring, and student grievances. Most universities require the department head to be a tenured faculty member with explicit fiduciary duties—a hard legal barrier.
Adoption barriers	claude-sonnet-5	5/5	Department head roles require formal institutional appointment, faculty governance approval, and accountability structures that legally and organizationally require a human in the position.
Cost vs. human wage	claude-haiku-4-5-20251001	1/5	Department head duties require legal accountability, institutional representation, and decision-making authority that cannot be delegated to AI systems. The cost of full human replacement via AI would exceed the human wage because the institution must retain a human principal and oversight mechanisms.
Cost vs. human wage	claude-sonnet-5	1/5	There is no AI substitute for the administrative leadership role, so no meaningful cost comparison for full task replacement exists; the human role remains necessary.
Technical feasibility today	claude-haiku-4-5-20251001	1/5	No deployed AI product reliably performs the role of department head in production. Tools exist for scheduling and document management, but none can independently execute the fiduciary and personnel responsibilities that define the role in any higher education institution today.
Technical feasibility today	claude-sonnet-5	1/5	No deployed product performs the role of a department head; AI tools at best support scheduling or document drafting within such a role, not the role itself.

Act as advisers to student organizations.

CI 0–11 · exposure 0 · augmentation 38 · importance 2.7/5 · click for rater detail

Panel ratings per dimension for this task
Dimension	Model	Rating	Rationale
Sector adoption velocity	claude-haiku-4-5-20251001	1/5	This task is inherently human-contact-dependent and has not seen any meaningful AI adoption in production; educational institutions remain strongly committed to human mentorship in this domain.
Sector adoption velocity	claude-sonnet-5	2/5	Higher education adopts AI tools slowly for interpersonal/mentorship roles, with adoption concentrated in administrative or research support rather than advising relationships.
Augmentation potential	claude-haiku-4-5-20251001	2/5	AI might assist advisers with administrative scheduling or document drafting, but the core advising work—listening, mentoring, and making judgment calls—remains human-centric, limiting augmentation value.
Augmentation potential	claude-sonnet-5	3/5	AI can help advisers with scheduling, drafting communications, budgeting suggestions, or event planning support, but the core advisory/mentorship relationship remains human-driven.
Task automatability	claude-haiku-4-5-20251001	1/5	Advising student organizations requires genuine relationship-building, mentorship, judgment about student developmental needs, and contextual decision-making that current AI systems cannot perform end-to-end. AI cannot replicate the trusted advisor role or provide meaningful personalized guidance.
Task automatability	claude-sonnet-5	1/5	Advising a student organization requires relational trust, mentorship, judgment about interpersonal/institutional dynamics, and physical presence at meetings/events that current AI cannot replicate end-to-end.
Adoption barriers	claude-haiku-4-5-20251001	5/5	Institutions legally and ethically require human advisers for student organizations; student welfare, pastoral responsibility, and institutional liability create hard barriers to substitution with AI systems.
Adoption barriers	claude-sonnet-5	4/5	Institutions typically require a faculty member to be officially designated as adviser for liability, funding, and oversight purposes, creating a strong organizational/administrative barrier to substitution.
Cost vs. human wage	claude-haiku-4-5-20251001	1/5	An AI system capable of performing this task reliably would require extensive customization and oversight, making it far more expensive than assigning the work to faculty or dedicated staff advisers.
Cost vs. human wage	claude-sonnet-5	2/5	There is essentially no AI product delivering this service, so cost comparison favors the human by default, though AI could cheaply handle minor scheduling/communication support.
Technical feasibility today	claude-haiku-4-5-20251001	1/5	No deployed products perform genuine student organization advising; the task requires human presence, accountability, and pastorally-inflected judgment that AI systems do not operationalize in production contexts.
Technical feasibility today	claude-sonnet-5	1/5	No deployed product performs the role of a student organization adviser; this remains firmly a human relational and administrative role.

Participate in campus and community events.

CI 0–5 · exposure 0 · augmentation 13 · importance 2.7/5 · click for rater detail

Panel ratings per dimension for this task
Dimension	Model	Rating	Rationale
Sector adoption velocity	claude-haiku-4-5-20251001	1/5	This task has zero adoption of AI automation because it is fundamentally tied to human presence and community relations, which are core institutional values in higher education.
Sector adoption velocity	claude-sonnet-5	1/5	This is an in-person, relational activity in higher education with no meaningful AI adoption trend for replacing physical attendance at events.
Augmentation potential	claude-haiku-4-5-20251001	1/5	AI offers no meaningful assistance in attending or participating in campus and community events; any supporting tools (scheduling, reminders) are orthogonal to the core participatory task itself.
Augmentation potential	claude-sonnet-5	2/5	AI could help draft talking points, event summaries, or follow-up communications, but offers minimal assistance to the core act of participating in person.
Task automatability	claude-haiku-4-5-20251001	1/5	Participating in campus and community events requires human presence, social interaction, relationship-building, and contextual judgment about which events to attend and how to engage meaningfully. AI cannot substitute for physical presence or authentic interpersonal participation.
Task automatability	claude-sonnet-5	1/5	Attending and participating in physical campus/community events requires human presence, social interaction, and relationship-building that AI cannot perform end-to-end.
Adoption barriers	claude-haiku-4-5-20251001	5/5	Campus and community engagement is intrinsically tied to institutional representation and human relationship-building; it is not something that can be delegated to non-human agents or outsourced without losing its essential value and purpose.
Adoption barriers	claude-sonnet-5	4/5	Community and campus engagement expects human representation and relationship-building tied to the faculty member's identity and role, creating strong organizational and social expectations against substitution.
Cost vs. human wage	claude-haiku-4-5-20251001	1/5	There is no AI-based alternative to human attendance at events, so cost comparison is moot; the task cannot be substituted by automation at any price.
Cost vs. human wage	claude-sonnet-5	1/5	There is no AI system that can replace a human's attendance and participation, so cost comparison favors the human by default.
Technical feasibility today	claude-haiku-4-5-20251001	1/5	No deployed AI system can physically attend events, engage in conversations, or represent an institution at community gatherings. This task fundamentally requires human presence and social agency.
Technical feasibility today	claude-sonnet-5	1/5	No deployed AI product substitutes for a person's physical/social participation in events; this is fundamentally a human presence task.

Supervise students' laboratory work.

CI 0–3 · exposure 0 · augmentation 25 · importance 4.1/5 · click for rater detail

Panel ratings per dimension for this task
Dimension	Model	Rating	Rationale
Sector adoption velocity	claude-haiku-4-5-20251001	1/5	Physics education remains heavily reliant on in-person instruction with minimal AI adoption for core pedagogical and safety functions; sector digitization is low for laboratory supervision itself.
Sector adoption velocity	claude-sonnet-5	2/5	Higher education is adopting AI for course content and grading, but physical lab supervision sees essentially no automation-driven change.
Augmentation potential	claude-haiku-4-5-20251001	2/5	AI tools can assist with data analysis or simulation support during lab work, but supervisory responsibilities—ensuring student safety, correcting technique, assessing understanding—remain fundamentally human-dependent with limited AI augmentation value.
Augmentation potential	claude-sonnet-5	2/5	AI can help prepare lab materials, quizzes, or pre-lab explanations, but offers little assistance to the moment-to-moment task of watching over students during hands-on lab work.
Task automatability	claude-haiku-4-5-20251001	1/5	Laboratory supervision requires real-time observation of student safety, hands-on problem-solving, and adaptive intervention—tasks that demand human presence and judgment that current AI cannot perform end-to-end in a physical lab environment.
Task automatability	claude-sonnet-5	1/5	Real-time physical supervision of students handling equipment, ensuring safety, and providing hands-on guidance cannot be performed by current AI systems, which lack physical presence and manipulation capability.
Adoption barriers	claude-haiku-4-5-20251001	5/5	Institutional accreditation standards, liability law, and safety regulations typically require a credentialed faculty member or qualified TA to be physically present during hazardous lab work; automation is legally and structurally blocked.
Adoption barriers	claude-sonnet-5	5/5	Safety regulations, liability for lab accidents, and institutional policy require a qualified human physically present to supervise students using equipment and chemicals.
Cost vs. human wage	claude-haiku-4-5-20251001	1/5	AI systems capable of monitoring physics labs would require significant hardware, integration, and continuous human oversight cost that would exceed the cost of employing a teaching assistant or instructor.
Cost vs. human wage	claude-sonnet-5	1/5	There is no AI substitute performing this physical, safety-critical supervisory role, so cost comparison favors the human by default.
Technical feasibility today	claude-haiku-4-5-20251001	1/5	No deployed product reliably performs unsupervised lab work supervision; while cameras and monitoring systems exist, they cannot substitute for a credentialed instructor's real-time safety oversight and pedagogical intervention.
Technical feasibility today	claude-sonnet-5	1/5	No deployed product provides in-person laboratory supervision; this remains entirely outside current AI product capabilities.

Serve on academic or administrative committees that deal with institutional policies, departmental matters, and academic issues.

CI 0–0 · exposure 0 · augmentation 38 · importance 3.1/5 · click for rater detail

Panel ratings per dimension for this task
Dimension	Model	Rating	Rationale
Sector adoption velocity	claude-haiku-4-5-20251001	1/5	Committee service is a fundamental governance function in academic institutions with no measurable AI displacement. Adoption in this domain remains negligible because the task is intrinsically human and organizational.
Sector adoption velocity	claude-sonnet-5	1/5	Higher education governance and committee work is a low-digitization, human-relationship-driven process with essentially no AI displacement occurring.
Augmentation potential	claude-haiku-4-5-20251001	2/5	AI could assist with minor tasks like drafting meeting agendas, summarizing prior discussions, or retrieving policy documents, but does not fundamentally enhance the human's core committee role of deliberation, judgment, and institutional decision-making.
Augmentation potential	claude-sonnet-5	3/5	AI can help summarize meeting materials, draft policy language, or prepare agendas, providing moderate assistance to committee members without replacing their participation.
Task automatability	claude-haiku-4-5-20251001	1/5	Committee service inherently requires human judgment on institutional policies, interpersonal negotiation, and consensus-building among stakeholders. AI cannot meaningfully participate in deliberations, voting, or the social-organizational functions that define committee work.
Task automatability	claude-sonnet-5	1/5	Committee service requires in-person deliberation, institutional politics awareness, judgment, and negotiation among stakeholders that AI cannot substitute for end-to-end.
Adoption barriers	claude-haiku-4-5-20251001	5/5	Institutional governance, committee participation, and decision-making on academic policy are legally and structurally tied to human faculty authority and institutional accountability. Faculty must personally serve and be liable for decisions.
Adoption barriers	claude-sonnet-5	5/5	Committee membership is tied to faculty status, governance structures, and institutional bylaws requiring human academic staff to serve and vote on policy matters.
Cost vs. human wage	claude-haiku-4-5-20251001	1/5	Committee service is a governance and institutional duty tied to employment and professional standing, not a discrete, cost-comparable task. The notion of replacing human committee participation with AI infrastructure does not yield a meaningful cost comparison.
Cost vs. human wage	claude-sonnet-5	1/5	There is no AI product replacing this function, so no meaningful cost comparison exists; the human role remains necessary regardless of cost.
Technical feasibility today	claude-haiku-4-5-20251001	1/5	No deployed AI system can serve on committees or represent institutional interests in governance bodies. While AI can draft documents or summarize issues, actually participating in committee meetings and decision-making remains outside the scope of current systems.
Technical feasibility today	claude-sonnet-5	1/5	No deployed product performs the role of a faculty committee member; this is inherently a human governance function.

Related occupations — Educational Instruction & Library

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.