Atmospheric, Earth, Marine, and Space Sciences Teachers, Postsecondary

25-1051.00
Median wage $103,170/yr9,900 employed (US)Rank #294 of 923 scored · top 32% by substitution

Teach courses in the physical sciences, except chemistry and physics. 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

Substitution34
Exposure32
Augmentation61

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

26 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

12%

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%33

panel mean rating 2.3/5 → substitution pressure 33/100

Technical feasibility todayw 20%31

panel mean rating 2.3/5 → substitution pressure 31/100

Cost vs. human wagew 15%38

panel mean rating 2.5/5 → substitution pressure 38/100

Adoption barriersw 20%inverted — strong barriers lower the score37

panel mean rating 3.5/5 (barrier strength) → substitution pressure 37/100

Sector adoption velocityw 10%30

panel mean rating 2.2/5 → substitution pressure 30/100

Task breakdown (26 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.

84

CI 74–95 · exposure 92 · augmentation 75 · importance 4.4/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510014/5Adoption of automated attendance and grading systems in postsecondary education is substantial and accelerating; most colleges and universities have deployed LMS platforms that automate these record-keeping tasks.
Sector adoption velocityclaude-sonnet-55/5Higher education has near-universal adoption of digital LMS and student information systems for these exact recordkeeping functions.
Augmentation potentialclaude-haiku-4-5-202510014/5AI and LMS tools substantially augment instructor productivity by automating routine data entry, flagging attendance patterns, and organizing records, freeing faculty to focus on teaching and student engagement while remaining in oversight.
Augmentation potentialclaude-sonnet-54/5AI-enhanced gradebooks and LMS analytics substantially reduce administrative burden and flag at-risk students, though instructors still verify and finalize records.
Task automatabilityclaude-haiku-4-5-202510014/5This task is highly automatable: attendance tracking, grade recording, and record maintenance can be performed by current learning management systems (Canvas, Blackboard, etc.) and AI agents with minimal human oversight, easily meeting the 50% time-saving threshold.
Task automatabilityclaude-sonnet-55/5Recordkeeping of attendance and grades is a structured data-entry/management task fully handled by existing LMS and gradebook software, often with automation and minimal human input beyond initial setup.
Adoption barriersclaude-haiku-4-5-202510014/5Significant barriers exist: institutional policy, IT governance requirements, FERPA compliance, integration with legacy systems, and organizational inertia around established record-keeping practices slow adoption even where technically feasible.
Adoption barriersclaude-sonnet-52/5Some institutional policies require faculty to review and certify final grades, but the underlying recordkeeping mechanics face no licensing or legal barriers to automation.
Cost vs. human wageclaude-haiku-4-5-202510015/5Modern LMS and automation tools cost a fraction of the loaded wage of a faculty member or administrator managing records manually; the per-task cost is orders of magnitude lower than human labor.
Cost vs. human wageclaude-sonnet-55/5Software-based recordkeeping costs a small licensing fee per institution versus substantial faculty/staff time, making automated systems drastically cheaper per record maintained.
Technical feasibility todayclaude-haiku-4-5-202510015/5Deployed LMS products reliably perform attendance tracking, grade recording, and record-keeping at scale in educational institutions today; these are mature, production-grade systems used daily in thousands of schools.
Technical feasibility todayclaude-sonnet-55/5Learning management systems (Canvas, Blackboard, PowerSchool, etc.) already perform automated grade calculation, attendance tracking, and record storage reliably at scale in production across universities.

Compile bibliographies of specialized materials for outside reading assignments.

83

CI 76–90 · exposure 83 · augmentation 88 · importance 2.8/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510013/5Higher education is gradually adopting AI tools for administrative and preparatory tasks, but bibliography compilation remains largely manual in most departments. Pockets of early adoption exist (large universities, STEM fields with heavy coursework), but sector-wide embedding is still in pilot phases.
Sector adoption velocityclaude-sonnet-53/5Higher education is adopting AI research tools steadily but unevenly, with individual faculty experimentation more common than institution-wide deployment for such tasks.
Augmentation potentialclaude-haiku-4-5-202510014/5AI significantly assists instructors by rapidly surfacing candidate sources, auto-formatting citations, and identifying thematic gaps in reading lists, allowing faculty to focus on pedagogical judgment and topical curation rather than clerical work. Productivity gains are substantial while humans retain editorial control.
Augmentation potentialclaude-sonnet-55/5AI substantially speeds up discovery and organization of specialized sources, letting instructors quickly draft and refine reading lists while still applying pedagogical judgment on final selections.
Task automatabilityclaude-haiku-4-5-202510014/5AI can reliably search academic databases, identify relevant peer-reviewed sources on specified topics, and format bibliographies in standard citation styles (APA, Chicago, MLA) with minimal human review. This saves 60–80% of the manual time spent locating and formatting sources, meeting the 50% threshold, though instructors typically verify topical appropriateness and currency.
Task automatabilityclaude-sonnet-55/5Compiling bibliographies on a topic is a well-defined literature-search task that current AI (search-augmented LLMs, citation tools) can perform end-to-end with major time savings, given a course topic and level.
Adoption barriersclaude-haiku-4-5-202510012/5No licensing or legal requirement mandates that a human instructor curate reading lists; most universities have already delegated or outsourced such work. Academic freedom norms and quality concerns create modest institutional friction, but no hard professional or regulatory barriers prevent automation.
Adoption barriersclaude-sonnet-51/5There is no licensing, regulatory, or liability barrier to using AI-generated bibliographies; faculty routinely delegate such prep work already.
Cost vs. human wageclaude-haiku-4-5-202510015/5AI inference cost for generating a bibliography is negligible (cents per task), while a human instructor manually compiling reading lists spends 30–90 minutes per assignment at loaded wage of $40–75/hour. Cost ratio favors AI by at least 100:1.
Cost vs. human wageclaude-sonnet-55/5Generating a bibliography via AI tools costs a fraction of a cent to a few dollars in compute versus the substantial hourly cost of a professor's time doing manual literature searches.
Technical feasibility todayclaude-haiku-4-5-202510014/5Deployed products (ChatGPT, Claude, specialized citation managers like Zotero with AI plugins, and academic search tools) can generate bibliographies and retrieve cited works reliably in production. Minor gaps exist in real-time access to the newest publications or niche specialized materials, but mainstream academic sources are handled well.
Technical feasibility todayclaude-sonnet-54/5Products like AI-powered research assistants, citation managers, and academic search tools (e.g., Elicit, Semantic Scholar recommendations, Perplexity) reliably generate topical reading lists today, though occasional inaccuracies or outdated sources require checking.

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

80

CI 76–84 · exposure 80 · augmentation 100 · importance 4.2/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510014/5Adoption is rapid in higher education, a digitized, information-sector environment. Many postsecondary institutions have seen widespread uptake of generative AI for syllabus and assignment drafting since late 2022, with pilots and production use now common among faculty in technical and science disciplines.
Sector adoption velocityclaude-sonnet-53/5Higher education adoption of AI for administrative/prep tasks is growing but uneven across departments and institutions, with many faculty still hesitant or using it informally rather than as standard workflow.
Augmentation potentialclaude-haiku-4-5-202510015/5AI substantially augments instructor productivity on this task: it drafts initial materials rapidly, freeing faculty to focus on content accuracy, disciplinary nuance, and alignment with learning outcomes rather than formatting and composition, while the instructor remains the final decision-maker on course design.
Augmentation potentialclaude-sonnet-55/5AI substantially speeds up drafting of syllabi, assignments, and handouts, letting instructors focus on customization, accuracy checks, and pedagogical judgment.
Task automatabilityclaude-haiku-4-5-202510014/5Current AI systems (GPT-4, Claude) can generate syllabi, homework assignments, and handouts at high quality with minimal human input. An instructor can prompt an LLM with learning objectives and receive 80–90% ready-to-use materials, achieving well over 50% time savings compared to manual composition, though final review and customization remain necessary.
Task automatabilityclaude-sonnet-54/5LLMs can draft syllabi, homework problem sets, and handouts effectively when given course topics and learning objectives, requiring mainly human review and customization.
Adoption barriersclaude-haiku-4-5-202510012/5While some institutions have governance policies around AI use in teaching and faculty may prefer human-authored materials for pedagogical reasons, there are no legal or licensing barriers preventing instructors from using AI to draft course materials. Organizational friction and preferences for human oversight exist but do not constitute hard substitution barriers.
Adoption barriersclaude-sonnet-52/5No licensing requirement blocks AI-assisted material creation, though some institutional policies or accreditation standards may require faculty authorship and review.
Cost vs. human wageclaude-haiku-4-5-202510015/5The per-task cost of AI inference and integration (API calls or subscriptions) is orders of magnitude cheaper than paying an instructor's loaded hourly wage to compose these materials manually, especially when bulk-generating materials for multiple courses or cohorts.
Cost vs. human wageclaude-sonnet-55/5Generating a first draft of a syllabus or homework set via AI costs a few cents to dollars compared to hours of faculty or TA time at academic wage rates.
Technical feasibility todayclaude-haiku-4-5-202510015/5Multiple deployed products (ChatGPT, Claude, specialized educational tools) reliably generate course materials in production use by educators today. Educational institutions and individual instructors are actively using these systems to draft syllabi and assignments with minimal error rates on routine tasks.
Technical feasibility todayclaude-sonnet-54/5Tools like ChatGPT, Claude, and specialized ed-tech platforms are already used by instructors to generate course materials in production, though discipline-specific accuracy (e.g., atmospheric science content) still needs verification.

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

67

CI 48–87 · exposure 70 · augmentation 75 · importance 4.2/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510014/5Postsecondary institutions (information and education sector) are adopting automated exam tools and AI grading at a rapidly accelerating pace, with widespread pilot programs and production deployment in many universities.
Sector adoption velocityclaude-sonnet-52/5Higher education adopts administrative AI tools unevenly and cautiously, with many departments still relying on manual grading, especially for specialized science courses.
Augmentation potentialclaude-haiku-4-5-202510014/5AI significantly enhances faculty productivity in exam creation (generating question banks, variations), administration (scheduling, proctoring support), and grading (flagging outliers, providing feedback drafts), while instructors retain control over content and standards.
Augmentation potentialclaude-sonnet-54/5AI substantially helps with generating question banks, formatting exams, and pre-grading objective items, letting instructors focus review time on complex or subjective responses.
Task automatabilityclaude-haiku-4-5-202510015/5Creating, administering, and grading examinations are all highly automatable with current AI. Multiple tools (learning management systems with AI modules, large language models for question generation, automated grading systems) can perform all three functions end-to-end with substantial time savings at comparable quality, meeting the ≥50% threshold.
Task automatabilityclaude-sonnet-53/5AI can draft exam questions and grade objective or short-answer responses reasonably well, but compiling exams aligned to specific course content and grading nuanced scientific reasoning still requires substantial instructor oversight.
Adoption barriersclaude-haiku-4-5-202510012/5While institutions and faculty may prefer human oversight and academic integrity concerns create some friction, there are no legal or licensing requirements mandating human creation or grading of exams; adoption is organizationally gated rather than regulated.
Adoption barriersclaude-sonnet-53/5No formal licensing requires a human to grade exams, but academic integrity policies, institutional grading standards, and instructor accountability create moderate friction against full automation.
Cost vs. human wageclaude-haiku-4-5-202510015/5AI-powered exam systems cost pennies per student per assessment versus instructor labor (at fully loaded $50–100+ per hour); the ratio is one to two orders of magnitude in AI's favor.
Cost vs. human wageclaude-sonnet-53/5AI tools can cut grading time significantly for standardized formats, but the need for human verification on technical content narrows the cost advantage to roughly comparable overall costs when oversight is included.
Technical feasibility todayclaude-haiku-4-5-202510014/5Deployed products (Canvas, Blackboard, Turnitin, ChatGPT-based plugins) reliably handle exam generation, administration, and grading in production at scale across postsecondary institutions, though some edge cases (complex subjective answers, nuanced scientific reasoning) require human review.
Technical feasibility todayclaude-sonnet-53/5Products like AI-assisted quiz generators and automated grading tools (e.g., Gradescope with AI features, LLM-based graders) are deployed in some institutions, but reliability for advanced STEM content with open-ended reasoning is inconsistent.

Write grant proposals to procure external research funding.

61

CI 54–67 · exposure 58 · augmentation 100 · importance 3.4/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510013/5Academic research is increasingly digitized, and AI writing tools are spreading in academia, but adoption of AI for grant writing specifically remains in the pilot and early-adoption phase; many institutions and researchers still view it with caution.
Sector adoption velocityclaude-sonnet-53/5Academic and research sectors show growing but uneven AI tool adoption for writing tasks; pilots and individual use are common but formal institutional workflows integrating AI into grant writing remain limited.
Augmentation potentialclaude-haiku-4-5-202510015/5AI significantly augments grant writing by rapidly generating first drafts, synthesizing literature, organizing budgets, and flagging structural gaps, allowing professors to focus on novel ideas and strategic framing rather than initial composition.
Augmentation potentialclaude-sonnet-55/5AI substantially speeds up drafting, editing, formatting, and literature synthesis for grant proposals, making it a strong productivity tool while the researcher retains ownership of ideas and final content.
Task automatabilityclaude-haiku-4-5-202510014/5AI can now draft substantial portions of grant proposals (background, methods, preliminary data synthesis, budget justification) with minimal human editing, meeting the 50% time-saving threshold. However, the creative framing of novel research contributions and the task of aligning proposals with funder priorities still benefit from human oversight.
Task automatabilityclaude-sonnet-53/5AI can draft substantial portions of grant proposals (background, literature review, methods framing) but requires deep domain expertise, original research ideas, and institutional knowledge that current AI cannot fully supply end-to-end at equal quality without heavy human revision.
Adoption barriersclaude-haiku-4-5-202510013/5Grant agencies expect human creativity, institutional accountability, and signature authority from faculty; there is no legal barrier to AI drafting, but institutional norms, funder review of originality, and the requirement for a human PI to certify claims create moderate friction.
Adoption barriersclaude-sonnet-52/5No formal licensing requirement to write grants, but institutional review, PI accountability, and funder trust in human-authored intellectual contributions create moderate friction against pure AI authorship.
Cost vs. human wageclaude-haiku-4-5-202510014/5AI inference and integration costs are negligible compared to the loaded wage of a professor or postdoc spending dozens of hours on a single proposal; even accounting for human review, the cost advantage is substantial.
Cost vs. human wageclaude-sonnet-54/5AI drafting tools cost very little (subscription fees) compared to the many hours a faculty member or grant writer would spend, though human review and refinement still add cost, keeping it just short of order-of-magnitude savings on the full task.
Technical feasibility todayclaude-haiku-4-5-202510013/5AI writing assistants and specialized tools like Elicit and Scholarcy are deployed and used by researchers, but they typically assist rather than replace the full workflow; grant writing remains partially manual and error-prone in critical sections (specific aims, innovation claims).
Technical feasibility todayclaude-sonnet-53/5Products like ChatGPT, Claude, and specialized grant-writing assistants are used in production by researchers today, but they still produce generic or inaccurate content requiring significant expert editing, especially for technical scientific content and funder-specific requirements.

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

51

CI 25–78 · exposure 58 · augmentation 88 · importance 3.8/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510014/5Academic institutions (especially STEM fields and information-intensive disciplines) are rapidly integrating AI tools into research workflows, with widespread pilot adoption of LLMs for drafting, analysis, and literature synthesis. However, formal policy frameworks and institutional resistance to full automation keep velocity high but not maximal compared to commercial sectors.
Sector adoption velocityclaude-sonnet-52/5Academic research adoption of AI tools is growing but remains uneven and cautious, with slow institutional and disciplinary norms around AI-assisted publication compared to fast-moving commercial sectors.
Augmentation potentialclaude-haiku-4-5-202510015/5AI dramatically augments researcher productivity: automated literature summarization, hypothesis generation, statistical analysis, visualization, and manuscript drafting allow human researchers to focus on experimental design, validation, and novelty. This transforms the research process while keeping the human as the authoritative decision-maker.
Augmentation potentialclaude-sonnet-54/5AI substantially aids literature review, statistical analysis, coding, manuscript drafting, and editing, meaningfully increasing researcher productivity while the scientist retains intellectual ownership and oversight.
Task automatabilityclaude-haiku-4-5-202510015/5Modern AI can automate substantial portions of the research workflow: literature review, data analysis, visualization, manuscript drafting, and formatting. While human conceptualization remains critical, AI can accelerate the end-to-end process—from data processing to submission-ready outputs—by well over 50%, meeting the time-saving threshold at comparable quality for many publishable research outputs.
Task automatabilityclaude-sonnet-52/5Original scientific research requiring novel field data collection, hypothesis generation, and expert judgment cannot be fully automated today, though AI can assist with literature review, data analysis, and drafting portions of the workflow.
Adoption barriersclaude-haiku-4-5-202510014/5Significant barriers exist: institutional reputation and authorship conventions require human researchers to vouch for novelty and integrity; funding agencies and tenure committees demand human intellectual ownership; and legal/ethical liability for false claims falls on the named human authors, not the AI system. These create strong friction against full automation.
Adoption barriersclaude-sonnet-54/5Academic publishing requires named human authorship, institutional accountability, ethical review, and peer credibility; journals and universities have strong norms and policies against AI as sole author or investigator.
Cost vs. human wageclaude-haiku-4-5-202510015/5The all-in cost of AI-assisted research (inference, integration, human oversight) is orders of magnitude lower than the fully-loaded salary of a postdoctoral researcher or junior faculty member conducting the same research; API costs and compute for a complete paper are typically <$100 versus ~$50k–$150k annual researcher salary for equivalent output.
Cost vs. human wageclaude-sonnet-52/5AI can cut costs for literature synthesis, coding, and drafting, but the core research (fieldwork, experiment design, peer-reviewed judgment) still requires expensive expert labor, keeping overall cost comparable to human-led research.
Technical feasibility todayclaude-haiku-4-5-202510014/5Deployed AI systems (large language models, statistical software with AI backends, citation management tools) demonstrably assist research writing, analysis, and publication workflows across academia. However, field-specific validation, peer review navigation, and novelty verification remain partially manual, limiting full end-to-end autonomy in production to a 4 rather than a 5.
Technical feasibility todayclaude-sonnet-52/5AI writing/research assistants and data-analysis tools are used by researchers, but no deployed product independently conducts original atmospheric/earth science research and produces publishable findings without substantial human direction.

Perform administrative duties, such as serving as department head.

50

CI 25–75 · exposure 58 · augmentation 75 · importance 3.3/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510013/5Higher education lags private-sector AI adoption; many universities use legacy systems and resist automation of governance roles. Pilot AI administrative assistance exists but widespread production deployment of AI for core department head functions remains uncommon relative to information and finance sectors.
Sector adoption velocityclaude-sonnet-52/5Higher education administration is a sector with relatively slow AI adoption for governance and leadership functions, though tools for scheduling and reporting are slowly being piloted.
Augmentation potentialclaude-haiku-4-5-202510015/5AI substantially augments department heads by automating scheduling, email filtering, document drafting, compliance reminders, and data synthesis for meetings, freeing time for strategic decisions and faculty relations while humans retain control and judgment on all consequential matters.
Augmentation potentialclaude-sonnet-53/5AI tools can help draft memos, summarize meeting notes, manage calendars, and analyze budget data, meaningfully assisting a department head's administrative workload.
Task automatabilityclaude-haiku-4-5-202510015/5AI can handle most administrative tasks end-to-end: scheduling, email management, document organization, budget tracking, meeting coordination, and report generation. Department head duties predominantly involve information processing and coordination rather than irreplaceable human judgment, allowing >50% time savings at equal quality with current AI systems.
Task automatabilityclaude-sonnet-52/5Administrative and leadership duties like serving as department head involve negotiation, personnel decisions, budgeting judgment, and institutional politics that require human relational and strategic capacity beyond current AI capability., only sub-tasks like scheduling or drafting reports are automatable.
Adoption barriersclaude-haiku-4-5-202510014/5Strong barriers exist: faculty governance traditions, union contracts, institutional policies requiring human sign-off on hiring/budget decisions, and legal/fiduciary liability for resource allocation typically mandate a licensed faculty member serve as department head and approve key decisions, preventing full substitution.
Adoption barriersclaude-sonnet-54/5Department head roles typically require institutional appointment, tenure status, and formal authority recognized by the university, creating strong organizational and credentialing barriers to automation.
Cost vs. human wageclaude-haiku-4-5-202510015/5AI infrastructure for administrative automation (LLMs, scheduling agents, document management) costs dollars per month, while replacing even fractional administrative duties of a department head (loaded cost >$100k/year) yields orders-of-magnitude cost savings.
Cost vs. human wageclaude-sonnet-52/5AI can cheaply handle small administrative sub-tasks, but the overall role requires substantial human oversight and judgment, so cost savings are limited to fragments of the task.
Technical feasibility todayclaude-haiku-4-5-202510014/5Production systems reliably perform core administrative functions (scheduling, email triage, document drafting, compliance tracking) in many universities and institutions today. Minor limitations exist in handling nuanced interpersonal conflicts or complex policy interpretation, but deployed AI handles the majority of department head administrative work reliably.
Technical feasibility todayclaude-sonnet-52/5AI products can assist with document drafting, scheduling, and data summarization, but no deployed product performs departmental leadership or administrative headship reliably in production.

Purchase and maintain equipment to support research projects.

49

CI 10–87 · exposure 45 · augmentation 50 · importance 3.2/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510014/5Higher-education procurement increasingly uses automated systems; universities have strong IT infrastructure and digitization, with broad adoption of enterprise procurement platforms. Implementation is common in production across research institutions.
Sector adoption velocityclaude-sonnet-51/5Academic research administration and equipment maintenance are low-digitization, physically-grounded activities with minimal AI agent adoption in production.
Augmentation potentialclaude-haiku-4-5-202510014/5AI-assisted procurement tools help faculty and administrators by surfacing supplier options, flagging cost overruns, recommending maintenance schedules, and automating routine reorders while leaving budget decisions and vendor relationships to humans. This substantially raises administrative productivity.
Augmentation potentialclaude-sonnet-52/5AI can assist with tracking inventory, comparing vendor prices, or drafting purchase orders, but cannot handle the physical maintenance or final purchasing decisions.
Task automatabilityclaude-haiku-4-5-202510015/5Equipment purchasing and maintenance scheduling can be largely automated today: AI can identify supplier options, compare costs, manage procurement workflows, schedule maintenance based on usage/calibration records, and generate purchase orders with significant time savings. Equipment tracking systems and automated reordering are well-established, meeting the ≥50% efficiency bar.
Task automatabilityclaude-sonnet-51/5This task involves physical procurement decisions, vendor negotiation, budget management, and hands-on equipment maintenance (e.g., lab instruments, sensors, field equipment) that AI cannot physically perform.
Adoption barriersclaude-haiku-4-5-202510012/5Academic institutions adopt e-procurement readily and few regulatory barriers exist for automated purchasing of research equipment. Light institutional friction (faculty preference for personal vendor relationships) and minor compliance requirements remain, but these are not hard barriers.
Adoption barriersclaude-sonnet-53/5While not licensed work, institutional procurement policies, budget authorization requirements, and physical maintenance needs create organizational friction against automation.
Cost vs. human wageclaude-haiku-4-5-202510015/5Automated procurement and maintenance management systems cost far less per transaction than dedicated human staff oversight; a single automated workflow handles multiple equipment purchases and schedules, making per-unit cost orders of magnitude lower than a full-time purchasing administrator's loaded wage.
Cost vs. human wageclaude-sonnet-51/5AI has no capability to execute physical purchasing or equipment maintenance, so there is no meaningful cost comparison—human labor is required.
Technical feasibility todayclaude-haiku-4-5-202510014/5Mature e-procurement platforms, inventory management systems, and maintenance scheduling tools perform these subtasks reliably in production across educational institutions. Some manual final approval and specialized equipment qualification remain, but most operational work is handled by deployed systems at scale.
Technical feasibility todayclaude-sonnet-51/5No deployed AI product purchases or physically maintains scientific research equipment; this remains entirely a human administrative and physical task.

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

39

CI 25–54 · exposure 38 · augmentation 63 · importance 4.2/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510012/5Higher education has adopted AI writing detectors and automated multiple-choice graders slowly and selectively, with widespread skepticism about accuracy and fairness. Full automation of grading in postsecondary science remains rare; most institutions still view human faculty grading as non-negotiable.
Sector adoption velocityclaude-sonnet-53/5Higher education has moderate AI adoption with growing use of grading assistants and plagiarism/feedback tools, but full-scale replacement of instructor grading remains uncommon and cautious.
Augmentation potentialclaude-haiku-4-5-202510013/5AI can meaningfully assist by identifying plagiarism flags, organizing submissions, flagging outlier responses for review, and providing preliminary rubric scoring on structured assignments. However, the core intellectual work of evaluating scientific reasoning and papers typically requires minimal AI input to remain effective.
Augmentation potentialclaude-sonnet-54/5AI substantially speeds up grading by drafting feedback, flagging errors, and checking objective components, letting instructors focus on final judgment and edge cases.
Task automatabilityclaude-haiku-4-5-202510012/5While AI can assist with grading objective assignments (quizzes, multiple choice), evaluating and grading papers and subjective class work requires nuanced judgment about scientific understanding, reasoning quality, and originality that current AI systems struggle with reliably. Significant human oversight and rework would be needed, falling well short of the 50% time-saving threshold.
Task automatabilityclaude-sonnet-53/5AI can grade objective assignments and provide draft feedback on written work reasonably well, but nuanced grading of scientific reasoning, lab reports, and original research papers still requires disciplinary expertise and judgment that current systems only partially replicate.
Adoption barriersclaude-haiku-4-5-202510014/5Faculty retain explicit responsibility for grade assignment and student evaluation; institutional policies, accreditation standards, and student appeals processes expect faculty judgment and sign-off. Liability for grade disputes and potential discrimination claims create strong friction against full automation of evaluation.
Adoption barriersclaude-sonnet-53/5No licensing requirement mandates human grading, but academic integrity, institutional grading policies, and instructor accountability create moderate friction against full automation.
Cost vs. human wageclaude-haiku-4-5-202510012/5AI grading tools require per-student licensing, integration setup, and substantial human verification of grades—particularly for papers requiring deep scientific judgment. The all-in cost per evaluated student work remains comparable to or higher than a faculty member's time on light grading, especially when accounting for liability and rework.
Cost vs. human wageclaude-sonnet-54/5Once integrated with rubrics and course content, AI grading assistance costs a small fraction of instructor or TA time per assignment, though initial setup and review add some cost.
Technical feasibility todayclaude-haiku-4-5-202510012/5Products like Canvas and Turnitin offer AI-assisted grading tools with plagiarism detection, but they perform poorly on open-ended scientific essays and complex conceptual work. These systems are limited to narrow rubric categories or pattern matching, not reliable evaluation of student understanding in postsecondary science contexts.
Technical feasibility todayclaude-sonnet-53/5AI-assisted grading tools and LLM-based feedback systems are deployed in some higher-ed contexts, but reliable, discipline-specific grading of postsecondary science coursework is not yet standard or fully trusted in production.

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

33

CI 23–44 · exposure 33 · augmentation 63 · importance 3.4/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510011/5Educational institutions, especially in purchasing and supply management, have lagged in AI adoption; procurement remains heavily manual and human-dependent in most postsecondary settings.
Sector adoption velocityclaude-sonnet-52/5Higher education administrative and procurement processes are generally slow to adopt AI tools compared to sectors like finance or tech, with most adoption still in early pilot stages for such logistical tasks.
Augmentation potentialclaude-haiku-4-5-202510013/5AI can assist by searching suppliers, comparing prices, and organizing inventory—useful aids that help humans work faster—but the core procurement decision and authorization remain human responsibilities.
Augmentation potentialclaude-sonnet-54/5AI tools can meaningfully assist by researching textbook options, comparing lab equipment specifications and prices, and drafting recommendations, significantly speeding up the research portion of this task even though final selection and procurement remain human-driven.
Task automatabilityclaude-haiku-4-5-202510012/5While AI can help identify and list needed materials and supplies, the task requires physical procurement decisions that involve vendor selection, budget constraints, and institutional preferences—activities that demand human judgment and authorization.
Task automatabilityclaude-sonnet-53/5AI can research, compare, and recommend textbooks and lab equipment options and even draft purchase orders, but final selection often requires judgment calls about curriculum fit, budget approval, and physical procurement logistics that need human execution.
Adoption barriersclaude-haiku-4-5-202510014/5Institutional procurement typically requires authorization by licensed personnel or budget-holders, and purchasing often involves legal contracts and liability considerations that mandate human oversight and signature.
Adoption barriersclaude-sonnet-52/5No licensing requirement, but institutional purchasing policies, budget authorization chains, and vendor relationships create moderate organizational friction that slows full automation.
Cost vs. human wageclaude-haiku-4-5-202510012/5AI could assist in research and cataloging, but human purchasing staff or educators must validate choices and process orders, making the combined cost comparable to or exceeding the salary cost of a staff member managing small procurement tasks.
Cost vs. human wageclaude-sonnet-52/5While AI-assisted research could cut some time, the overall task still requires human interaction with vendors, purchasing systems, and budget approvals, so total cost savings versus a faculty member or admin doing this directly are modest.
Technical feasibility todayclaude-haiku-4-5-202510012/5No deployed product reliably handles end-to-end procurement for educational institutions; existing e-procurement systems require human decision-making at critical junctures and lack the contextual reasoning needed to independently select appropriate lab equipment.
Technical feasibility todayclaude-sonnet-52/5There are no deployed products specifically automating academic materials procurement for postsecondary science instructors; general-purpose AI assistants can help with research and comparison but institutions still rely on manual ordering systems and human decision-making.

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

29

CI 25–34 · exposure 25 · augmentation 63 · importance 4.2/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510012/5Higher education has historically slower AI adoption than other sectors, and curriculum planning remains a core faculty prerogative with limited evidence of widespread AI agent deployment in production. Pilots exist but production substitution is minimal.
Sector adoption velocityclaude-sonnet-52/5Higher education adoption of AI for curriculum design is still in early pilot stages, with slow institutional processes limiting broad deployment despite growing interest in AI tools for teaching support.
Augmentation potentialclaude-haiku-4-5-202510013/5AI can meaningfully assist instructors by generating content drafts, suggesting activity structures, and organizing reference materials, moderately raising productivity while faculty retain full decision authority over pedagogical direction and institutional fit.
Augmentation potentialclaude-sonnet-54/5AI tools substantially help faculty brainstorm content, draft materials, summarize research, and suggest revisions, meaningfully speeding up parts of the curriculum development process while faculty retain final judgment.
Task automatabilityclaude-haiku-4-5-202510012/5AI can assist in generating course content drafts and organizing materials, but curriculum planning requires pedagogical judgment, institutional context awareness, and alignment with learning outcomes that demand human expertise. The task involves significant creative and evaluative components beyond 50% time savings.
Task automatabilityclaude-sonnet-52/5AI can draft syllabi or suggest content but the holistic evaluation and revision of curricula requires disciplinary judgment, institutional context, and accreditation alignment that current systems cannot fully replicate end-to-end.
Adoption barriersclaude-haiku-4-5-202510014/5Curriculum design at postsecondary institutions typically requires faculty governance, accreditation alignment, and disciplinary expertise. Institutional policies and professional norms strongly favor human faculty authority over course design, creating significant organizational and governance barriers.
Adoption barriersclaude-sonnet-53/5No licensing barrier exists, but curriculum design is typically tied to faculty governance, accreditation standards, and departmental review processes that create organizational friction.
Cost vs. human wageclaude-haiku-4-5-202510013/5AI-powered content generation and organization tools cost relatively little per use, but the overhead of instructor review, integration with institutional systems, and validation of educational quality roughly balances against faculty time for initial planning.
Cost vs. human wageclaude-sonnet-52/5While AI can cheaply generate draft content, the human oversight, expert review, and institutional approval processes needed keep overall cost savings modest relative to faculty time.
Technical feasibility todayclaude-haiku-4-5-202510012/5While AI tools can generate outlines and content summaries, no deployed product reliably handles end-to-end curriculum planning with the contextual sensitivity required for postsecondary science education. Production systems exist for content generation but lack the integrated pedagogical reasoning needed.
Technical feasibility todayclaude-sonnet-52/5AI writing assistants are used to help draft course materials, but no deployed product autonomously plans and revises entire postsecondary curricula reliably in production.

Answer questions from the public and media.

28

CI 23–34 · exposure 25 · augmentation 63 · importance 2.8/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510011/5Academic institutions have not meaningfully adopted AI systems to independently handle media and public inquiries from faculty. Adoption remains negligible; universities continue to rely on faculty, communications offices, and traditional protocols rather than AI-mediated responses.
Sector adoption velocityclaude-sonnet-52/5Higher education adoption of AI for public-facing communication is still slow and cautious, with most faculty using AI informally rather than as an institutional practice for media relations.
Augmentation potentialclaude-haiku-4-5-202510013/5AI can assist faculty by drafting initial responses, organizing talking points, or rapidly retrieving relevant background material, raising efficiency in preparation. However, the faculty member remains the necessary final voice; AI functions as a research and drafting aid rather than a transformative productivity multiplier.
Augmentation potentialclaude-sonnet-54/5AI tools can efficiently help faculty draft responses, summarize research for lay audiences, and prepare talking points before public or media engagements, meaningfully boosting productivity while the human remains the face of the response.
Task automatabilityclaude-haiku-4-5-202510012/5While AI can generate plausible answers to straightforward factual questions about atmospheric or earth sciences, teaching contexts require nuanced explanation, authority representation, and handling of follow-up questions that demand real expertise. Current systems cannot reliably substitute for a credentialed instructor without significant human oversight, falling short of the 50% time-saving threshold for end-to-end performance.
Task automatabilityclaude-sonnet-52/5AI can draft answers to common science questions but a postsecondary instructor's public/media engagement often requires institutional authority, personal expertise, and nuanced framing that current systems can't fully replace end-to-end.
Adoption barriersclaude-haiku-4-5-202510014/5Universities have institutional reputability concerns and media-relations policies; public statements on scientific matters carry authority and liability implications. Faculty credentialing and institutional accountability requirements create substantial friction against pure automation; answers typically require human expert sign-off and authorization.
Adoption barriersclaude-sonnet-53/5There's no formal licensing barrier, but institutional expectations, reputational stakes, and the value of named expert authority create meaningful friction against full AI substitution.
Cost vs. human wageclaude-haiku-4-5-202510012/5Implementing AI systems for public/media response would still require a faculty member to review, authorize, and manage responses, plus infrastructure and liability oversight. The cost of full-stack deployment plus human accountability remains close to or exceeds the cost of the faculty member answering directly.
Cost vs. human wageclaude-sonnet-53/5AI-assisted drafting is cheap, but the task still requires the professor's time for review, attribution, and credibility, keeping overall cost roughly comparable to having the human handle it directly with light AI support.
Technical feasibility todayclaude-haiku-4-5-202510012/5Large language models can produce coherent responses to science questions, but deployed systems lack institutional credibility, make factual errors, and cannot reliably represent an academic institution or handle sensitive media inquiries. No mature product does this task independently in production at university scale.
Technical feasibility todayclaude-sonnet-52/5Chatbots and search tools can answer general science questions reasonably well, but no deployed product substitutes for a named faculty expert responding to media inquiries or public correspondence in this professional capacity.

Initiate, facilitate, and moderate classroom discussions.

28

CI 25–30 · exposure 25 · augmentation 75 · importance 4.0/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510012/5Adoption remains slow and experimental; most higher-education institutions use AI for administrative or content support, not for replacing or automating the core teaching function of classroom moderation, which remains deeply rooted in human expertise and institutional conservatism.
Sector adoption velocityclaude-sonnet-52/5Higher education adopts AI unevenly; while some instructors use AI for course prep, actual classroom facilitation remains predominantly human-led with slow institutional change.
Augmentation potentialclaude-haiku-4-5-202510014/5AI can meaningfully assist instructors by generating discussion prompts, real-time transcription and summarization, identifying unanswered questions, and providing student engagement metrics, significantly enhancing preparation and follow-up while the instructor remains the active moderator.
Augmentation potentialclaude-sonnet-54/5AI can generate discussion questions, prompts, and follow-up ideas, and help instructors prepare more engaging and structured discussions, meaningfully boosting preparation productivity.
Task automatabilityclaude-haiku-4-5-202510012/5Current AI can partially support discussion by generating prompts, summarizing points, and identifying gaps, but cannot authentically replace the real-time facilitation, dynamic adaptation, relationship-building, and social presence required to moderate a live classroom discussion where students interact with an instructor.
Task automatabilityclaude-sonnet-52/5Leading and moderating live classroom discussion requires real-time responsiveness, reading student engagement, and adaptive pedagogy that current AI cannot reliably replicate end-to-end in a physical classroom setting.
Adoption barriersclaude-haiku-4-5-202510014/5Strong barriers exist: institutional norms and accreditation require direct instructor-student contact in higher education, faculty governance protects teaching roles, and institutional culture emphasizes the irreplaceability of live classroom dialogue for learning outcomes and student mentorship.
Adoption barriersclaude-sonnet-53/5No formal licensing bars AI from suggesting discussion prompts, but institutional norms, accreditation expectations, and student preference for human interaction create real friction against full delegation.
Cost vs. human wageclaude-haiku-4-5-202510012/5AI tools that support discussion preparation or post-hoc analysis are inexpensive, but a full facilitation system would require substantial integration and human oversight costs comparable to or exceeding the marginal cost of instructor time.
Cost vs. human wageclaude-sonnet-52/5Human instructors are already paid to teach as part of broader duties, and any AI facilitation tool would need significant integration and oversight, offering little cost advantage over the instructor already present.
Technical feasibility todayclaude-haiku-4-5-202510012/5Some teaching assistants and chatbots exist for discussion support and Q&A, but no deployed system reliably manages the full cycle of initiating engagement, moderating diverse student voices, and redirecting unproductive conversations in a real classroom setting with consistent quality.
Technical feasibility todayclaude-sonnet-52/5AI chatbots and discussion-board moderation tools exist for online forums, but no deployed product autonomously facilitates live in-person postsecondary classroom discussion at scale.

Participate in student recruitment, registration, and placement activities.

28

CI 25–30 · exposure 25 · augmentation 50 · importance 3.0/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510012/5Higher education has historically lagged in automation adoption due to cultural emphasis on student-faculty relationships and institutional conservatism. While some universities pilot chatbots for registration, placement and recruitment remain heavily human-driven with limited AI integration in production.
Sector adoption velocityclaude-sonnet-52/5Higher education is a relatively slow-adopting sector for AI in core academic/administrative functions like recruitment and placement, with pilots more common than full production deployment.
Augmentation potentialclaude-haiku-4-5-202510013/5AI can assist faculty by automating scheduling, generating prospect lists, summarizing student applications, and flagging placement matches, meaningfully reducing administrative load. However, the human judgment required for mentorship and placement guidance limits the upside of augmentation.
Augmentation potentialclaude-sonnet-53/5AI tools can meaningfully assist with drafting recruitment materials, managing communications, and pre-screening applications, improving efficiency while faculty retain final judgment.
Task automatabilityclaude-haiku-4-5-202510012/5Only narrow parts of this task—such as sending templated recruitment emails or parsing registration forms—can be automated with current AI. The core activities of student recruitment, registration support, and placement require human relationship-building, judgment about fit, and personalized guidance that AI cannot replicate end-to-end at equal quality.
Task automatabilityclaude-sonnet-52/5This task involves relational, evaluative, and institutional decision-making (interviews, advising, admissions judgment calls) that current AI cannot fully replace, though some sub-tasks like scheduling or initial screening could be automated.
Adoption barriersclaude-haiku-4-5-202510014/5Educational institutions face high organizational friction around student experience and accountability; students and parents expect human advising; and placement decisions carry reputational and liability weight that institutions are reluctant to delegate to AI without human sign-off. Accreditation bodies also expect human-led advising.
Adoption barriersclaude-sonnet-53/5No strict licensing requirement, but institutional policy, accreditation standards, and the expectation of faculty involvement in admissions/placement decisions create moderate organizational friction against full automation.
Cost vs. human wageclaude-haiku-4-5-202510012/5Chatbot and registration automation have low inference costs, but full integration into university systems, oversight, and the fallback human review needed for placement decisions make the all-in cost approach the cost of a human staff member without replacing them entirely.
Cost vs. human wageclaude-sonnet-52/5AI can cheaply handle communications and initial screening, but the faculty member's personal involvement (interviews, relationship-building, subjective evaluation) still requires human time, keeping overall cost savings modest.
Technical feasibility todayclaude-haiku-4-5-202510012/5While chatbots can handle basic FAQ responses and form processing, no mature product reliably performs the full spectrum of recruitment, registration counseling, and placement guidance that higher education requires. Deployed systems exist only for narrow subtasks like application screening.
Technical feasibility todayclaude-sonnet-52/5Some products exist for chatbot-based recruitment outreach or application screening, but comprehensive student recruitment/placement involving faculty judgment and personal interaction is not reliably handled by deployed AI systems today.

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

27

CI 16–38 · exposure 17 · augmentation 63 · importance 4.0/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510012/5Higher education remains a laggard in AI adoption for core pedagogical and research duties. While some faculty use AI-generated summaries or literature tools as aids, there is no evidence of systematic replacement of this professional development task in the academy.
Sector adoption velocityclaude-sonnet-53/5Academic and research sectors have moderately adopted AI tools for literature review and summarization, though full task automation lags due to the social/professional nature of the work.
Augmentation potentialclaude-haiku-4-5-202510013/5AI tools for literature discovery, paper summarization, and conference abstract browsing can usefully assist faculty in filtering and organizing information. However, the augmentation is partial—AI cannot replicate the insight, discussion quality, or professional relationships that define staying meaningfully current.
Augmentation potentialclaude-sonnet-54/5AI literature search, summarization, and alert tools significantly speed up staying current with publications, meaningfully augmenting this task even though human networking and conference presence remain essential.
Task automatabilityclaude-haiku-4-5-202510011/5This task fundamentally requires human judgment about which developments matter, synthesis of diverse information sources, and relationship-building through collegial discussion. While AI can help surface relevant papers or summarize literature, the core activity—staying informed and prioritizing what's important—depends on a professional's situated expertise and judgment.
Task automatabilityclaude-sonnet-52/5AI can help surface and summarize literature, but the underlying task of genuinely staying current via reading, professional dialogue, and conference participation requires ongoing human engagement and judgment that current systems cannot fully replace.
Adoption barriersclaude-haiku-4-5-202510014/5Professional development and currency in one's field are expected responsibilities embedded in academic employment, peer recognition, and institutional norms. Institutions and disciplines expect faculty to maintain expertise through direct engagement; replacement by AI would undermine the legitimacy and effectiveness of the role.
Adoption barriersclaude-sonnet-52/5No licensing requirement blocks AI assistance, but professional norms, tacit knowledge exchange, and networking value keep this fundamentally a human activity within academia.
Cost vs. human wageclaude-haiku-4-5-202510012/5Automating parts of literature review via AI summaries is inexpensive, but the human must still invest time in filtering, discussing, and integrating insights. The cost savings from partial automation are modest relative to the faculty member's loaded wage, and cannot eliminate the person-hours required.
Cost vs. human wageclaude-sonnet-52/5AI tools for literature scanning are cheap, but they only cover a fraction of the task; the human still needs to attend conferences, network, and synthesize, so overall cost savings are modest.
Technical feasibility todayclaude-haiku-4-5-202510012/5AI can retrieve and summarize recent papers or conference abstracts automatically, but cannot replace the human acts of reading with critical evaluation, having substantive conversations with colleagues, or attending conferences. No deployed product performs the full end-to-end task of staying genuinely abreast in a field.
Technical feasibility todayclaude-sonnet-52/5Products like literature-summarization tools and research assistants exist and are used, but no deployed system autonomously performs the full scope of tracking field developments through networking and conference attendance.

Prepare and deliver lectures to undergraduate or graduate students on topics such as structural geology, micrometeorology, and atmospheric thermodynamics.

23

CI 16–30 · exposure 25 · augmentation 63 · importance 4.4/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510012/5Higher education institutions have shown slow and cautious adoption of AI for core instructional tasks; most experimentation remains in supplementary roles (automated grading, tutoring assistants) rather than lecture delivery. Cultural and regulatory resistance to automation of primary teaching functions remains strong across the sector.
Sector adoption velocityclaude-sonnet-52/5Higher education adoption of AI for teaching is still largely pilot-stage, with slow institutional change processes and faculty caution limiting deep production use for actual lecture delivery.
Augmentation potentialclaude-haiku-4-5-202510013/5AI can meaningfully assist lecturers by generating presentation slides, creating practice problems, summarizing research papers, and drafting explanations of complex concepts, improving preparation efficiency. However, augmentation is limited to pre- and post-class tasks rather than enhancing real-time pedagogical delivery or student interaction.
Augmentation potentialclaude-sonnet-54/5AI tools substantially help postsecondary teachers draft lecture outlines, generate visuals, explain complex concepts, and create practice problems, meaningfully boosting preparation productivity while the instructor still delivers and adapts the lecture.
Task automatabilityclaude-haiku-4-5-202510012/5While AI can generate lecture outlines and slides on scientific topics, end-to-end lecture delivery requires real-time student engagement, adaptive explanation, and interactive problem-solving that current AI systems cannot reliably perform. Some lecture preparation components (outline drafting, visual material generation) could be partially automated, but core delivery remains largely human-dependent.
Task automatabilityclaude-sonnet-52/5AI can help draft lecture content and explanations, but live delivery, adapting to student questions, and pedagogical judgment for specialized graduate-level topics still require substantial human involvement, falling short of full end-to-end automation with equal quality.
Adoption barriersclaude-haiku-4-5-202510015/5Postsecondary instruction is heavily regulated by accreditation bodies, institutional policies, and professional standards that require qualified human faculty with subject-matter credentials to bear responsibility for student learning and degree-granting authority. Legal and contractual frameworks mandate faculty involvement in course delivery.
Adoption barriersclaude-sonnet-53/5No licensing requirement mandates a human deliver lectures, but accreditation norms, student expectations, and tenure-based academic structures create moderate institutional friction against full replacement.
Cost vs. human wageclaude-haiku-4-5-202510011/5The full-loaded cost of a faculty member delivering specialized postsecondary instruction (including curriculum design, office hours, assessment) far exceeds current AI inference costs, and substitution would require significant institutional overhead to maintain educational quality and student outcomes that institutions must guarantee.
Cost vs. human wageclaude-sonnet-52/5AI can cheaply generate slide content and drafts, but developing accurate, discipline-specific lecture content plus in-person delivery and interaction still requires expert oversight, keeping costs comparable to or only modestly below a professor's time for this task.
Technical feasibility todayclaude-haiku-4-5-202510012/5AI lecture generation tools and content systems exist in limited forms, but no mature product reliably replaces live undergraduate/graduate instruction with comparable learning outcomes. Current systems lack the ability to respond to student questions spontaneously, adjust explanations based on comprehension signals, or handle the dynamic classroom environment that defines this task.
Technical feasibility todayclaude-sonnet-52/5Some university pilots use AI-generated lecture materials or recorded AI tutors, but no mature product reliably delivers full graduate-level lectures on specialized topics like micrometeorology in production at scale.

Review papers or serve on editorial boards for scientific journals, and review grant proposals for federal agencies.

23

CI 20–25 · exposure 20 · augmentation 50 · importance 2.8/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510012/5Academic and federal grant systems move slowly on structural changes; adoption of AI-driven peer review remains experimental and cautious. Most journals and agencies continue to rely on traditional human review, with AI playing only a supporting role in screening.
Sector adoption velocityclaude-sonnet-52/5Academic publishing and grant review are notoriously slow to adopt new technology due to entrenched peer-review norms, ethics concerns, and institutional inertia, despite some experimentation with AI-assisted screening.
Augmentation potentialclaude-haiku-4-5-202510013/5AI can usefully assist reviewers by summarizing papers, flagging potential issues, checking citations, and organizing data—reducing routine work. However, the core assessment still depends on human expert judgment, so augmentation is valuable but bounded.
Augmentation potentialclaude-sonnet-53/5AI can help draft summaries, check statistical methods, flag plagiarism or missing citations, and streamline literature comparison, meaningfully aiding reviewers without replacing their judgment.
Task automatabilityclaude-haiku-4-5-202510012/5AI can assist with preliminary screening and formatting checks on papers and proposals, but cannot reliably perform the core judgment work of peer review—assessing novelty, methodological soundness, and significance requires domain expertise, contextual knowledge, and accountability that current systems cannot fully replicate. Significant human oversight would remain necessary.
Task automatabilityclaude-sonnet-52/5Peer review and grant evaluation require deep domain expertise, judgment about novelty, methodological rigor, and scientific significance that current AI cannot reliably replicate end-to-end. AI can assist with parts (checking references, summarizing) but not fully substitute the reviewer's judgment.
Adoption barriersclaude-haiku-4-5-202510014/5Peer review and grant evaluation carry legal, reputational, and accountability requirements; journals and federal agencies require named human reviewers to take responsibility for recommendations. Regulatory and institutional norms strongly mandate human expert involvement and sign-off.
Adoption barriersclaude-sonnet-54/5Editorial boards and funding agencies require named, credentialed experts to sign off on reviews; conflict-of-interest rules, accountability, and reputational trust structures create strong institutional and quasi-regulatory barriers to full automation.
Cost vs. human wageclaude-haiku-4-5-202510012/5A full peer review or grant review cycle involving human experts incurs significant cost in time and expertise. AI tools reduce ancillary work but do not eliminate the need for expert reviewers, so overall cost savings remain modest compared to expert compensation.
Cost vs. human wageclaude-sonnet-52/5AI inference is cheap, but reliable scientific judgment requires expert oversight and correction, making the effective all-in cost of an AI-assisted review process still comparable to or more than uncompensated/volunteer human review given liability and quality risks.
Technical feasibility todayclaude-haiku-4-5-202510012/5While AI tools exist for manuscript analysis and summary generation, no deployed system performs independent peer review or grant evaluation at production quality. Tools like text analysis and similarity detection support the process but do not replace the expert decision-making required by journals and federal agencies.
Technical feasibility todayclaude-sonnet-51/5No deployed product performs full peer review or grant assessment autonomously in production; journals and agencies still require human expert reviewers, with AI tools used only as ancillary aids (plagiarism checks, formatting).

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

21

CI 7–34 · exposure 13 · augmentation 63 · importance 3.6/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510012/5Adoption of AI for student advising remains limited to narrow pilot use (information lookup, scheduling support) in higher education. The personalized, high-stakes nature of career advising and institutional resistance to removing human advisors from this role means deployment velocity is slow and adoption remains nascent.
Sector adoption velocityclaude-sonnet-52/5Higher education adopts AI advising tools slowly and unevenly, with pilots more common than full production deployment for personalized career counseling.
Augmentation potentialclaude-haiku-4-5-202510013/5AI can assist human advisors by surfacing relevant program information, labor-market data, prerequisite requirements, and career pathways, helping advisors work more efficiently. However, the task's core—understanding the student and synthesizing a personalized path—remains substantially human, so augmentation is meaningful but not transformative.
Augmentation potentialclaude-sonnet-54/5AI can help advisors research career paths, degree requirements, and job market data, meaningfully speeding up preparation for advising sessions.
Task automatabilityclaude-haiku-4-5-202510011/5Advising students on academic and vocational curricula and career issues requires deep personal judgment, understanding of individual student circumstances, long-term career trajectory planning, and contextual knowledge that current AI cannot reliably replicate end-to-end. While AI can provide information, the holistic synthesis with student goals, institutional constraints, and mentorship—core to effective advising—remains firmly human-dependent.
Task automatabilityclaude-sonnet-52/5Advising involves personalized judgment, institutional knowledge, and relationship-building that current AI cannot fully replicate end-to-end, though AI can assist with information lookup and drafting.atable
Adoption barriersclaude-haiku-4-5-202510014/5Institutions typically treat student advising as a licensed or credentialed responsibility (faculty or trained advisors), and there is strong organizational and liability preference for human accountability. Students and families expect human judgment and mentorship on major life decisions, creating both regulatory and cultural friction against full automation.
Adoption barriersclaude-sonnet-53/5No licensing requirement for academic advising, but institutional norms, accreditation expectations, and student preference for human mentorship create moderate friction.
Cost vs. human wageclaude-haiku-4-5-202510011/5The cost-per-advising-session from an AI system plus oversight would likely exceed the cost of a trained human advisor, particularly when accounting for the need to review, correct, and verify AI recommendations that could affect students' academic and career trajectories. Trust and error-cost asymmetry make this economically unfavorable.
Cost vs. human wageclaude-sonnet-53/5AI tools are cheap to run but human faculty advising is often bundled into existing teaching duties, so marginal cost savings are moderate rather than dramatic.
Technical feasibility todayclaude-haiku-4-5-202510011/5No deployed product reliably performs the full scope of academic and career advising in production today. Chatbots can answer factual questions about programs or careers, but real advising requires sustained relationship-building, nuanced understanding of a student's aspirations and constraints, and accountability for outcomes that current AI systems do not demonstrably handle at scale.
Technical feasibility todayclaude-sonnet-52/5Chatbot advising tools exist in some universities but are narrow (scheduling, FAQs) and not trusted for substantive career/curriculum guidance at scale.

Provide professional consulting services to government or industry.

16

CI 6–25 · exposure 13 · augmentation 63 · importance 2.4/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510012/5While some sectors are experimenting with AI-assisted research for consulting, actual replacement or autonomous consulting delivery is rare in production. Most adoption remains limited to internal efficiency gains rather than client-facing service displacement.
Sector adoption velocityclaude-sonnet-52/5Academic and specialized scientific consulting sectors adopt AI tools slowly compared to fast-moving information/finance sectors, with pilots more common than production deployment.
Augmentation potentialclaude-haiku-4-5-202510013/5AI can assist consulting experts by accelerating literature review, data analysis, scenario modeling, and draft report generation, meaningfully improving productivity on analytical components while the human retains full responsibility for client engagement and final recommendations.
Augmentation potentialclaude-sonnet-54/5AI can significantly help with literature review, data analysis, report drafting, and scenario modeling, substantially boosting the expert's productivity while they retain final judgment and client relationship.
Task automatabilityclaude-haiku-4-5-202510011/5Consulting services require deep domain expertise, client relationship management, and judgment about complex real-world problems. Current AI cannot establish client trust, negotiate contract terms, or take responsibility for strategic recommendations—core elements of professional consulting.
Task automatabilityclaude-sonnet-52/5Consulting requires synthesizing domain expertise, contextual judgment, and client-specific negotiation that current AI cannot fully replace, though it can assist with research and analysis components.dim.
Adoption barriersclaude-haiku-4-5-202510015/5Professional consulting is heavily protected by licensing requirements (professional engineers, environmental consultants in many jurisdictions), legal liability for recommendations, regulatory compliance, and industry/client expectations that a qualified human expert must author and sign off on deliverables.
Adoption barriersclaude-sonnet-54/5Government and industry consulting often requires named credentialed experts, professional liability, and accountability for advice given, creating strong barriers to full AI substitution.
Cost vs. human wageclaude-haiku-4-5-202510012/5While AI can reduce research and document preparation costs, the loaded human wage for a consulting expert (often $150–300+/hour) includes irreplaceable credibility and liability assumption that AI cannot replicate, keeping overall cost advantage modest.
Cost vs. human wageclaude-sonnet-52/5While AI can cheaply produce draft analyses, the human expert's credibility, liability, and judgment remain necessary, so all-in cost including oversight is not dramatically lower than the human consultant.
Technical feasibility todayclaude-haiku-4-5-202510011/5No deployed product performs professional consulting end-to-end in production. AI tools can assist with research or drafting, but the client-facing, advisory, and accountability aspects remain entirely dependent on human experts.
Technical feasibility todayclaude-sonnet-52/5No deployed products autonomously deliver professional consulting engagements in atmospheric/earth sciences; AI is used as a research aid but not as a substitute consultant of record.

Collaborate with colleagues to address teaching and research issues.

8

CI 0–16 · exposure 0 · augmentation 38 · importance 3.5/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510011/5Even in high-tech academic environments, AI-driven autonomous collaboration remains research-stage. Institutions have not adopted systems to automate collegial deliberation on research and curriculum matters.
Sector adoption velocityclaude-sonnet-52/5Higher education is a moderate-to-slow adopter of AI for interpersonal/collaborative academic functions, with pilots more common in research support than in collegial decision-making.
Augmentation potentialclaude-haiku-4-5-202510012/5AI can marginally assist by summarizing prior discussions or generating reference material before meetings, but does not meaningfully augment the core intellectual and social work of reaching agreement with peers.
Augmentation potentialclaude-sonnet-53/5AI can assist by summarizing research literature, drafting meeting notes, or synthesizing curriculum data to inform discussions, but the core collaborative exchange remains human-driven.
Task automatabilityclaude-haiku-4-5-202510011/5Meaningful collaboration on teaching and research issues requires interpersonal negotiation, consensus-building, and context-dependent judgment about academic priorities. Current AI cannot autonomously participate in genuine collegial problem-solving.
Task automatabilityclaude-sonnet-51/5Collaborative professional dialogue among colleagues on teaching approaches and research direction requires relational trust, shared context, and real-time judgment that AI cannot substitute for end-to-end.
Adoption barriersclaude-haiku-4-5-202510015/5Academic collaboration is fundamentally a human social and intellectual process embedded in institutional governance. Colleagues must interact directly to align on research direction and teaching strategy; no substitution by AI is expected or feasible.
Adoption barriersclaude-sonnet-53/5No formal licensing barrier exists, but strong organizational and cultural norms around academic collegiality and shared governance create real friction against AI substitution.
Cost vs. human wageclaude-haiku-4-5-202510012/5AI could assist with some preparatory tasks (literature synthesis, meeting notes), but cannot replace the human time spent in actual collaboration; assistance cost is modest compared to the loaded academic salary.
Cost vs. human wageclaude-sonnet-51/5There is no AI substitute performing this interpersonal task, so cost comparison is not meaningful; humans remain the only viable option.
Technical feasibility todayclaude-haiku-4-5-202510011/5No deployed product reliably performs autonomous participation in academic collaboration meetings or substantive research consensus-building. AI tools lack the social understanding and authority to engage as a collegial peer.
Technical feasibility todayclaude-sonnet-51/5No deployed product performs collegial collaboration on teaching/research issues; AI tools may support communication but cannot conduct the collaboration itself.

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

6

CI 0–13 · exposure 5 · augmentation 50 · importance 3.4/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510011/5Higher education remains highly resistant to automating human mentorship and supervision; adoption of AI in academic supervision is minimal and limited to back-office tasks, not the core supervisory relationship.
Sector adoption velocityclaude-sonnet-52/5Higher education adopts AI tools slowly for core mentorship functions, though usage of AI for research assistance and drafting feedback is growing modestly.
Augmentation potentialclaude-haiku-4-5-202510013/5AI can assist with administrative scheduling, draft feedback generation, and progress tracking, but the core mentoring and research oversight remain fundamentally human activities where AI plays a supporting role.
Augmentation potentialclaude-sonnet-53/5AI can help supervisors review drafts, generate feedback, track student progress, and suggest research directions, meaningfully aiding but not replacing supervisory judgment.
Task automatabilityclaude-haiku-4-5-202510011/5Supervising teaching, internships, and research work requires ongoing human judgment about student progress, interpersonal feedback, mentorship decisions, and adaptive guidance based on individual needs—elements that current AI cannot handle end-to-end at equal quality.
Task automatabilityclaude-sonnet-51/5Supervising students' research and teaching requires ongoing personalized mentorship, judgment calls, and relationship-building that current AI cannot autonomously replicate end-to-end.
Adoption barriersclaude-haiku-4-5-202510015/5Academic institutions require faculty to hold formal responsibility for student supervision, mentorship quality, and research oversight; this duty of care and liability is legally and contractually assigned to licensed faculty members.
Adoption barriersclaude-sonnet-54/5Academic institutions require credentialed faculty to formally supervise theses, grades, and research ethics compliance, creating strong institutional and accreditation barriers.
Cost vs. human wageclaude-haiku-4-5-202510011/5The loaded cost of faculty supervision is already embedded in academic employment; AI tools for administrative assistance would supplement rather than replace the core supervision function, making direct cost comparison unfavorable for full automation.
Cost vs. human wageclaude-sonnet-51/5Since AI cannot perform the core supervisory role, there is no viable substitute cost comparison—human faculty time remains necessary and irreplaceable.
Technical feasibility todayclaude-haiku-4-5-202510012/5While AI can assist with scheduling and administrative tracking, no deployed system reliably supervises the full scope of mentoring, evaluation, and adaptive guidance that characterizes academic supervision in production settings.
Technical feasibility todayclaude-sonnet-51/5No deployed product supervises student research or internships; AI tools at best assist with feedback on drafts, not act as a supervisor of record.

Act as advisers to student organizations.

6

CI 5–7 · exposure 0 · augmentation 25 · importance 2.8/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510011/5Education, especially postsecondary mentoring, remains highly resistant to automation; adoption of AI for advising roles is minimal and limited to administrative information-retrieval support, not decision-making.
Sector adoption velocityclaude-sonnet-52/5Higher education adopts AI tools for teaching support, but advisory/mentorship roles for student groups remain largely untouched by AI adoption trends.
Augmentation potentialclaude-haiku-4-5-202510012/5AI might assist advisors by surfacing student records or organizational data, but the core task—counsel, motivation, and judgment—remains fundamentally human, with only marginal productivity uplift possible.
Augmentation potentialclaude-sonnet-52/5AI can help with scheduling, drafting communications, or budget tracking for the organization, but core advising, mentorship, and relationship functions see minimal AI augmentation.
Task automatabilityclaude-haiku-4-5-202510011/5Advising student organizations requires relationship-building, nuanced understanding of individual student contexts, mentoring, and real-time judgment—activities that demand human presence and emotional intelligence that current AI cannot replicate end-to-end.
Task automatabilityclaude-sonnet-51/5Advising student organizations requires relationship-building, mentorship, institutional knowledge, and in-person judgment that AI cannot replicate or substantially replace.
Adoption barriersclaude-haiku-4-5-202510014/5Student advising involves fiduciary responsibility, institutional liability for student outcomes, and an organizational expectation that faculty provide this service; universities would face governance and accreditation friction attempting to substitute AI for human advisors.
Adoption barriersclaude-sonnet-54/5Institutional policies typically require a designated faculty/staff advisor for liability, mentorship, and administrative accountability reasons, creating strong organizational and quasi-regulatory barriers.
Cost vs. human wageclaude-haiku-4-5-202510011/5The human advisor's embedded institutional knowledge, availability for ongoing support, and ability to navigate organizational dynamics make replacement by AI economically implausible and organizationally inferior.
Cost vs. human wageclaude-sonnet-51/5There is no viable AI substitute performing this role, so cost comparison favors the human by default since AI cannot deliver the output at all.
Technical feasibility todayclaude-haiku-4-5-202510011/5No deployed AI product reliably performs the advisory relationship role for student organizations; this task requires sustained engagement, trust-building, and context-awareness that exceeds current capabilities.
Technical feasibility todayclaude-sonnet-51/5No deployed AI products serve as student organization advisors; this is a fundamentally human relational and mentorship role.

Maintain regularly scheduled office hours to advise and assist students.

6

CI 0–11 · exposure 0 · augmentation 38 · importance 3.6/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510011/5Despite digitization of higher education, office hours remain a human touchstone; adoption of AI for direct student advising in academic settings remains minimal, with institutions prioritizing faculty presence and one-on-one relationships as core to the educational mission.
Sector adoption velocityclaude-sonnet-52/5Higher education adopts AI slowly for direct student advising due to institutional inertia, though AI tutoring tools are increasingly piloted alongside, not instead of, office hours.
Augmentation potentialclaude-haiku-4-5-202510012/5AI could assist a professor by pre-screening student questions, drafting responses to frequently asked questions, or organizing student records, but would marginally enhance rather than transform the core advising conversation, which depends on human presence and judgment.
Augmentation potentialclaude-sonnet-53/5AI can help by answering routine student questions, drafting responses, or triaging inquiries before office hours, improving efficiency without replacing the personal advising interaction.
Task automatabilityclaude-haiku-4-5-202510011/5This task fundamentally requires real-time, personalized interaction with individual students to address their unique questions, concerns, and academic progress. AI cannot reliably replicate the adaptive, empathetic guidance and nuanced judgment needed for effective advising across diverse student situations.
Task automatabilityclaude-sonnet-51/5This task inherently requires a human presence and relational engagement with students; AI cannot fulfill the scheduled, in-person/synchronous advising role itself.the core value is human availability and mentorship.
Adoption barriersclaude-haiku-4-5-202510015/5Faculty are contractually obligated to provide office hours; students expect and are often entitled to speak with the actual instructor; institutional policy and accreditation standards typically require direct faculty-student contact for advising and support, creating strong legal and regulatory barriers to substitution.
Adoption barriersclaude-sonnet-54/5Institutional norms, accreditation expectations, and student-faculty relationship requirements create strong organizational and quasi-regulatory barriers to replacing this with AI.
Cost vs. human wageclaude-haiku-4-5-202510011/5The cost of deploying, maintaining, and overseeing an AI system to handle office hours advising—combined with the liability and reputational risk of failures—would exceed the cost of a faculty member's hourly office time, which is already embedded in salary.
Cost vs. human wageclaude-sonnet-52/5While AI chat tools are cheap for answering FAQs, the task specifically requires the human's scheduled availability, so cost comparison for the actual task is not favorable to AI replacement.
Technical feasibility todayclaude-haiku-4-5-202510011/5While chatbots can provide factual information, no deployed product reliably substitutes for the human advisor role in postsecondary office hours, where relationship-building, mentoring, and contextual problem-solving are essential and expected by students and institutions.
Technical feasibility todayclaude-sonnet-51/5No deployed product substitutes for a professor's physical or live office-hour presence; chatbots can supplement but not replace this scheduled human interaction.

Participate in campus and community events.

5

CI 5–5 · exposure 0 · augmentation 25 · importance 2.8/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510011/5This task is embedded in institutional culture and faculty role expectations; there is no trajectory toward AI adoption because the task is fundamentally about human presence and community relationships.
Sector adoption velocityclaude-sonnet-51/5Higher education community engagement is a low-digitization, relationship-driven activity with essentially no AI adoption trend for physical attendance tasks.
Augmentation potentialclaude-haiku-4-5-202510012/5AI might assist with logistical planning, event scheduling, or promotional materials for such events, but the core task of participating requires human attendance and engagement, limiting meaningful augmentation.
Augmentation potentialclaude-sonnet-52/5AI can help with scheduling, event promotion materials, or talking points beforehand, but offers minimal assistance during the actual act of participating.
Task automatabilityclaude-haiku-4-5-202510011/5Participating in campus and community events requires human presence, relationship-building, and genuine engagement with attendees. AI cannot meaningfully replace the interactive, social, and relational components that define participation in these events.
Task automatabilityclaude-sonnet-51/5Physical/social participation in campus and community events requires human presence, relationship-building, and embodied engagement that AI cannot perform.'
Adoption barriersclaude-haiku-4-5-202510014/5Institutional and professional norms strongly expect faculty to participate in campus life and community engagement. There is an implicit but firm organizational expectation that a human educator will be the one present and engaged at these events.
Adoption barriersclaude-sonnet-54/5Participation is inherently tied to physical presence, institutional representation, and social/professional norms expecting a human faculty member, creating strong non-regulatory but practical barriers.
Cost vs. human wageclaude-haiku-4-5-202510011/5AI systems have no cost advantage here because the task cannot be automated; a human teacher must be physically and socially present, making the human cost unavoidable and irreplaceable.
Cost vs. human wageclaude-sonnet-51/5There is no AI substitute performing this task, so cost comparison favors the human by default since AI cannot deliver the output at all.
Technical feasibility todayclaude-haiku-4-5-202510011/5No deployed product can autonomously participate in campus and community events in the manner required of a postsecondary teacher. This task fundamentally requires human presence and authentic engagement.
Technical feasibility todayclaude-sonnet-51/5No deployed AI product attends or participates in in-person events on behalf of a person; this remains entirely outside current product capability.

Supervise laboratory work and field work.

3

CI 0–5 · exposure 0 · augmentation 25 · importance 4.2/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510011/5Educational institutions are slow to adopt automation in core instructional supervision roles, and there is strong cultural and liability-driven preference for human presence during hands-on learning. Adoption remains minimal.
Sector adoption velocityclaude-sonnet-51/5Higher education physical science instruction is a low-digitization, in-person-intensive sector with minimal AI displacement of supervisory duties.
Augmentation potentialclaude-haiku-4-5-202510012/5AI could provide limited assistance (e.g., automated hazard detection via video analytics, data logging during experiments) but does not meaningfully enhance instructor productivity in the core task of real-time supervision and student guidance.
Augmentation potentialclaude-sonnet-52/5AI can help prepare lab materials, checklists, or safety protocols, but offers little direct assistance during live supervision itself.
Task automatabilityclaude-haiku-4-5-202510011/5Supervising laboratory and field work requires real-time presence, safety oversight, student interaction, and adaptive decision-making that current AI systems cannot perform end-to-end. Automation would need to physically monitor students, assess hazards, and provide immediate guidance—capabilities that do not exist today.
Task automatabilityclaude-sonnet-51/5Supervising laboratory and field work requires physical presence, real-time safety oversight, and hands-on demonstration that current AI cannot perform end-to-end.'
Adoption barriersclaude-haiku-4-5-202510015/5Educational institutions have legal and ethical obligations to ensure student safety during lab and field work; liability frameworks generally require a qualified human instructor to be present and responsible. Professional standards and institutional policies create hard barriers to substitution.
Adoption barriersclaude-sonnet-54/5Safety liability, institutional policy, and often accreditation/certification requirements mean a qualified human must be present to supervise students in labs and field settings.
Cost vs. human wageclaude-haiku-4-5-202510011/5Even if partial automation were possible, the cost of required hardware (cameras, sensors, monitoring systems), integration, and human oversight would likely exceed or match the cost of direct instructor supervision, particularly given low-cost human labor in academic settings.
Cost vs. human wageclaude-sonnet-51/5There is no AI substitute performing this physical supervisory role, so cost comparison favors the human by default.
Technical feasibility todayclaude-haiku-4-5-202510011/5No deployed product reliably supervises laboratory or field work. Such supervision demands embodied presence, real-time safety assessment, and dynamic student interaction that exceeds what current AI systems can do in production educational settings.
Technical feasibility todayclaude-sonnet-51/5No deployed AI product supervises physical lab or field activities involving students; this remains outside current product capabilities.

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

0

CI 0–0 · exposure 0 · augmentation 38 · importance 3.3/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510011/5Academic governance remains highly resistant to automation and fundamentally human-centered. There is no sector movement to delegate committee service to AI systems.
Sector adoption velocityclaude-sonnet-51/5Higher education governance is slow-moving and highly traditional, with no observed trend of AI systems taking on committee membership roles.
Augmentation potentialclaude-haiku-4-5-202510012/5AI could assist with meeting preparation, agenda synthesis, or document analysis, but the core deliberative and voting functions cannot be augmented by delegating them to AI; only pre-work and information synthesis benefit marginally.
Augmentation potentialclaude-sonnet-53/5AI can help draft policy documents, summarize meeting materials, analyze data for committee decisions, or prepare agendas, offering moderate productivity support.
Task automatabilityclaude-haiku-4-5-202510011/5Committee service requires judgment about institutional policy, interpersonal negotiation, and nuanced decision-making in real organizational contexts. Current AI cannot participate meaningfully in deliberative processes or cast votes on complex institutional matters.
Task automatabilityclaude-sonnet-51/5Committee service requires in-person deliberation, political judgment, relationship navigation, and institutional context that AI cannot perform in place of a human representative.
Adoption barriersclaude-haiku-4-5-202510015/5Hard legal and institutional barriers exist: only faculty members with standing can serve on committees; governance requires human decision-makers accountable to the institution. Committee participation is a licensed and authorized institutional role.
Adoption barriersclaude-sonnet-55/5Committee membership is tied to institutional governance structures, faculty status, and often bylaws/tenure requirements that mandate human faculty representation and accountability.
Cost vs. human wageclaude-haiku-4-5-202510011/5Committee service is a human responsibility with legal and fiduciary dimensions. The cost of AI attempting to replicate this (monitoring, oversight, potential liability) would exceed the uncompensated or minimally compensated time faculty already contribute.
Cost vs. human wageclaude-sonnet-51/5There is no viable AI substitute performing this role, so cost comparison favors the human by default; AI cannot replace the representative function.
Technical feasibility todayclaude-haiku-4-5-202510011/5No deployed product can autonomously serve on academic committees; this task fundamentally requires human deliberation, accountability, and institutional authority that cannot be delegated to AI systems.
Technical feasibility todayclaude-sonnet-51/5No deployed product substitutes for a faculty member's participation and vote on academic or administrative committees today.

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.