Climate Change Policy Analysts
19-2041.01Median wage $82,220/yr89,250 employed (US)Rank #195 of 923 scored · top 21% by substitution
Research and analyze policy developments related to climate change. Make climate-related recommendations for actions such as legislation, awareness campaigns, or fundraising approaches.
Sub-scores
0–100 · band = confidence interval from rater disagreement
Substitution39
Exposure35
Augmentation81
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
14 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
7%
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%37
panel mean rating 2.5/5 → substitution pressure 37/100
Cost vs. human wagew 15%47
panel mean rating 2.9/5 → substitution pressure 47/100
Adoption barriersw 20%inverted — strong barriers lower the score46
panel mean rating 3.1/5 (barrier strength) → substitution pressure 46/100
Sector adoption velocityw 10%34
panel mean rating 2.4/5 → substitution pressure 34/100
Task breakdown (14 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.
Gather and review climate-related studies from government agencies, research laboratories, and other organizations.
71CI 67–75 · exposure 70 · augmentation 100 · importance 4.1/5 · click for rater detail
Gather and review climate-related studies from government agencies, research laboratories, and other organizations.
71| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 4/5 | Policy research, academic institutions, and government agencies are actively adopting AI-powered literature review tools, research assistants, and document management systems. Academic and policy sectors show strong digitization and rapid uptake of generative AI for research tasks. |
| Sector adoption velocity | claude-sonnet-5 | 3/5 | Policy research and government-adjacent analytical work show moderate AI adoption—pilots and tools increasingly used for literature synthesis, but formal production workflows lag behind sectors like finance or tech. |
| Augmentation potential | claude-haiku-4-5-20251001 | 5/5 | AI significantly enhances analyst productivity by rapidly surfacing relevant studies, generating summaries, identifying key findings, and cross-referencing sources—allowing the human analyst to focus on critical evaluation and synthesis rather than mechanical search and organization. |
| Augmentation potential | claude-sonnet-5 | 5/5 | AI substantially accelerates the gathering, filtering, and summarizing of large study volumes, letting analysts focus on interpretation and policy implications while remaining in the loop. |
| Task automatability | claude-haiku-4-5-20251001 | 4/5 | Current AI systems can gather, filter, and summarize climate-related studies from digital repositories and web sources at scale with significant time savings. While full end-to-end automation with quality control requires some human oversight of selection criteria and validation, AI can easily achieve >50% time reduction on literature review, search, and initial classification tasks that dominate this work. |
| Task automatability | claude-sonnet-5 | 4/5 | AI can efficiently search, summarize, and synthesize large volumes of climate-related studies and reports, saving significant time versus manual literature review, though final judgment on relevance and quality still requires human oversight. |
| Adoption barriers | claude-haiku-4-5-20251001 | 2/5 | There are minimal legal or regulatory barriers to automating literature gathering and review. Organizations may prefer human judgment for final selection and interpretation, but the gatekeeping task itself (searching, retrieving, filtering studies) faces no licensing or hard authorization requirements. |
| Adoption barriers | claude-sonnet-5 | 2/5 | No licensing requirement for gathering/reviewing studies, though some organizational trust and accuracy-verification friction exists given the policy-relevance of the material. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 4/5 | AI-driven literature gathering, filtering, and summarization costs (via API calls, SaaS platforms, or deployed agents) are orders of magnitude cheaper than hiring analysts to manually search databases, download papers, and produce initial summaries. A human analyst's loaded cost easily exceeds the per-task AI cost. |
| Cost vs. human wage | claude-sonnet-5 | 4/5 | AI-assisted search and summarization tools are dramatically cheaper per document processed than analyst hours, even after accounting for verification overhead. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 4/5 | Mature products (academic search engines with AI backends, literature review platforms, document summarization tools, and web scraping systems) reliably perform large-scale gathering and initial review of scientific documents in production. Deployed systems can access government databases, institutional repositories, and research platforms with minimal error on retrieval and classification. |
| Technical feasibility today | claude-sonnet-5 | 3/5 | Deployed tools (AI research assistants, RAG-based literature review tools, summarization products) exist and are used for document gathering and synthesis, but accuracy on niche technical/scientific sources and proper sourcing still has material error rates. |
Propose new or modified policies involving use of traditional and alternative fuels, transportation of goods, and other factors relating to climate and climate change.
49CI 25–72 · exposure 53 · augmentation 88 · importance 4.5/5 · click for rater detail
Propose new or modified policies involving use of traditional and alternative fuels, transportation of goods, and other factors relating to climate and climate change.
49| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 2/5 | Policy-making institutions, government agencies, and NGOs have historically been slow to adopt automation and AI in core policy functions. While digital tools are used for research, actual AI-driven policy proposal generation remains in pilot phase rather than deployed production across the sector. |
| Sector adoption velocity | claude-sonnet-5 | 2/5 | Policy analysis in government and environmental organizations is a professional-services-adjacent field but adoption is slower than finance or tech, with AI used mainly for research support rather than proposal generation. |
| Augmentation potential | claude-haiku-4-5-20251001 | 5/5 | AI assistants significantly boost productivity for climate analysts by rapidly synthesizing research, generating policy alternatives, and helping structure complex multi-factor proposals while analysts retain full strategic control and expertise over final direction and stakeholder alignment. |
| Augmentation potential | claude-sonnet-5 | 4/5 | AI can meaningfully assist by summarizing research, drafting sections, modeling scenarios, and comparing precedent policies, substantially speeding up the analyst's workflow while they retain final judgment. |
| Task automatability | claude-haiku-4-5-20251001 | 5/5 | Large language models can synthesize climate policy research, generate coherent policy proposals that integrate fuel, transportation, and climate factors, and produce complete draft policies at scale. Given access to existing policy frameworks and climate data, current AI systems can meet the 50% time-saving bar for policy ideation and proposal drafting. |
| Task automatability | claude-sonnet-5 | 2/5 | Drafting policy proposals requires synthesizing scientific, economic, and political considerations with judgment calls that current AI can support but not independently perform at equal quality; only portions like research summarization and drafting can be automated with significant time savings. |
| Adoption barriers | claude-haiku-4-5-20251001 | 4/5 | Adoption is constrained by accountability requirements, stakeholder legitimacy concerns, and the political and legal nature of policy work. Government agencies and NGOs face institutional friction and liability questions around AI-generated policy that humans must ultimately validate and sign off on, creating a strong human-in-the-loop barrier. |
| Adoption barriers | claude-sonnet-5 | 3/5 | No formal licensing requirement blocks AI use, but institutional accountability, political sensitivity, and the need for defensible expert judgment create real organizational friction against fully automated proposals. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 5/5 | The inference cost of an LLM-based policy drafting system is orders of magnitude cheaper than the fully-loaded salary of a climate policy analyst or team required to generate comparable proposals from scratch. |
| Cost vs. human wage | claude-sonnet-5 | 2/5 | While AI can cheaply generate draft text, the expert review, stakeholder analysis, and validation needed to produce a usable policy proposal keep the loaded cost comparable to or only modestly cheaper than a skilled analyst's time. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 3/5 | AI tools exist and are used for policy research assistance and proposal generation, but deployed systems typically require significant human expert review, domain validation, and stakeholder iteration before adoption. No mature product fully owns policy proposal generation end-to-end in production government or NGO settings without heavy human oversight. |
| Technical feasibility today | claude-sonnet-5 | 2/5 | AI writing tools and research assistants exist but there is no deployed product that reliably generates full, defensible climate policy proposals in production without heavy human oversight and revision. |
Provide analytical support for policy briefs related to renewable energy, energy efficiency, or climate change.
49CI 39–59 · exposure 42 · augmentation 88 · importance 4.8/5 · click for rater detail
Provide analytical support for policy briefs related to renewable energy, energy efficiency, or climate change.
49| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 2/5 | Government and NGO sectors adopting AI slowly; policy organizations remain conservative about AI in core analytical work due to accountability and accuracy concerns. Pilot projects exist but production displacement is minimal. |
| Sector adoption velocity | claude-sonnet-5 | 3/5 | Policy analysis and think-tank/government sectors are adopting AI tools for research and drafting at a moderate pace, with pilots more common than fully embedded production workflows. |
| Augmentation potential | claude-haiku-4-5-20251001 | 4/5 | AI excels at accelerating literature review, data aggregation, and generating structured drafts that analysts refine. Tools that retrieve climate data, model outputs, and policy documents substantially raise analyst productivity while human judgment guides conclusions. |
| Augmentation potential | claude-sonnet-5 | 5/5 | AI substantially speeds up literature review, data synthesis, and drafting for policy briefs, letting analysts focus on judgment, framing, and stakeholder-specific recommendations. |
| Task automatability | claude-haiku-4-5-20251001 | 2/5 | AI can draft sections of policy briefs and synthesize data on renewable energy trends, but policy analysis requires contextual judgment, stakeholder consideration, and normative framing that AI cannot reliably execute end-to-end. A human analyst must substantially shape the argument and conclusions. |
| Task automatability | claude-sonnet-5 | 3/5 | AI can draft literature summaries, synthesize data, and produce first-draft analysis, but integrating current policy context, verifying data accuracy, and tailoring to specific political stakeholders still requires substantial human judgment and revision. |
| Adoption barriers | claude-haiku-4-5-20251001 | 3/5 | Policy briefs often require attribution to credentialed analysts and face internal review processes; some agencies have guidance on AI use in policy work. Organizational norms favor human authorship, but no hard legal barrier prevents AI-assisted drafting. |
| Adoption barriers | claude-sonnet-5 | 2/5 | No licensing requirement blocks AI use in drafting analytical content, though organizational review processes and accountability for policy accuracy create moderate friction. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 3/5 | AI inference and integration costs are falling, but policy briefs still require human expert review, fact-checking, and rewriting. The all-in cost (labor + oversight) is roughly comparable to a junior analyst's wage for moderate-complexity briefs. |
| Cost vs. human wage | claude-sonnet-5 | 4/5 | AI-assisted drafting and research synthesis is much cheaper per unit output than a human analyst's time, though oversight and fact-checking costs reduce the savings somewhat. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 3/5 | Tools like LLMs can assist with research synthesis, report generation, and data summarization; some organizations use them for draft briefs. However, deployment for policy analysis is limited by the need for domain expertise, accuracy verification, and alignment with political/organizational positions. |
| Technical feasibility today | claude-sonnet-5 | 3/5 | LLM-based research and writing assistants are used in policy shops today for drafting and summarization, but reliability on technical accuracy, current data, and nuanced political framing remains inconsistent without human review. |
Review existing policies or legislation to identify environmental impacts.
47CI 37–56 · exposure 42 · augmentation 75 · importance 4.1/5 · click for rater detail
Review existing policies or legislation to identify environmental impacts.
47| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 2/5 | Government agencies and environmental bodies move slowly on AI adoption due to regulatory risk and institutional inertia; pilot programs exist but production deployment of autonomous policy review remains limited. |
| Sector adoption velocity | claude-sonnet-5 | 2/5 | Government and policy analysis sectors are generally slower adopters of AI tools compared to finance or tech, with pilots more common than full production deployment for this type of analysis. |
| Augmentation potential | claude-haiku-4-5-20251001 | 4/5 | AI excels at rapidly surfacing relevant clauses, prior environmental impact data, and cross-policy contradictions, significantly accelerating a human analyst's review and synthesis of environmental implications. |
| Augmentation potential | claude-sonnet-5 | 4/5 | AI substantially aids analysts by quickly surfacing relevant provisions, summarizing lengthy legislation, and flagging potential environmental impacts for further human evaluation. |
| Task automatability | claude-haiku-4-5-20251001 | 2/5 | AI can extract and summarize key environmental impacts from policy documents, but cannot reliably interpret ambiguous legislative intent, weigh competing environmental trade-offs, or identify context-dependent implications that require domain expertise and policy judgment. |
| Task automatability | claude-sonnet-5 | 3/5 | AI can rapidly summarize and flag environmental provisions across large volumes of legislative text, but nuanced assessment of environmental impact often requires jurisdiction-specific context, causal reasoning, and integration with scientific data that current systems handle only partially. |
| Adoption barriers | claude-haiku-4-5-20251001 | 4/5 | Policy review often requires professional credentials, institutional authority, and sign-off on regulatory compliance; organizations are legally and reputationally liable for inaccurate environmental impact assessment, creating strong gatekeeping against full automation. |
| Adoption barriers | claude-sonnet-5 | 2/5 | No licensing requirement mandates a human to perform this specific review step, though final policy recommendations in government contexts often require analyst sign-off and accountability, creating moderate friction. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 4/5 | AI document processing and summarization is now substantially cheaper than human policy review labor, reducing per-analysis cost by 50–70%, though human expertise for final judgment adds overhead. |
| Cost vs. human wage | claude-sonnet-5 | 4/5 | Once configured, AI can process and summarize large policy corpora far more cheaply than analyst hours, though initial setup, domain tuning, and validation add cost overhead. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 3/5 | Document analysis and impact extraction tools exist and work in production for many policy review tasks, but they require significant human oversight and struggle with nuanced multi-stakeholder impacts or novel regulatory frameworks. |
| Technical feasibility today | claude-sonnet-5 | 3/5 | Legal/policy AI tools and LLM-based document analysis products exist and are used for policy review, but reliable environmental-impact-specific extraction with low error rates is still narrow and often requires human verification. |
Write reports or academic papers to communicate findings of climate-related studies.
44CI 34–55 · exposure 42 · augmentation 75 · importance 3.8/5 · click for rater detail
Write reports or academic papers to communicate findings of climate-related studies.
44| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 2/5 | Academic and policy institutions are cautious adopters of AI for research writing due to validation concerns, publication ethics, and the reputational cost of AI-generated errors in high-stakes climate discourse. Adoption remains largely pilot or assistive rather than mainstream displacement. |
| Sector adoption velocity | claude-sonnet-5 | 3/5 | Policy research and academic sectors are adopting AI writing tools at a moderate pace, with pilots and partial integration common but full production-scale replacement still limited by accuracy concerns. |
| Augmentation potential | claude-haiku-4-5-20251001 | 4/5 | AI substantially assists by drafting outlines, generating literature summaries, editing prose, and accelerating revision cycles. A climate analyst using AI for structuring arguments and iterative writing can increase output quality and speed while retaining expert control over findings and interpretations. |
| Augmentation potential | claude-sonnet-5 | 4/5 | AI substantially assists with literature summarization, drafting, editing, and structuring reports, meaningfully boosting analyst productivity while the human retains responsibility for accuracy and interpretation. |
| Task automatability | claude-haiku-4-5-20251001 | 2/5 | AI can draft sections, organize arguments, and generate readable prose, but climate research reports require expert judgment on methodology critique, novel synthesis of domain-specific findings, and credible interpretation of complex models—tasks where AI often conflates plausibility with accuracy. Significant human oversight and rewriting are needed. |
| Task automatability | claude-sonnet-5 | 3/5 | AI can draft substantial portions of reports summarizing data and findings, but synthesizing novel policy-relevant analysis, ensuring accuracy of claims, and framing conclusions with domain judgment still requires significant human effort, so only partial time savings are achievable. |
| Adoption barriers | claude-haiku-4-5-20251001 | 4/5 | Climate policy reports often feed into regulatory, legislative, or international frameworks where institutional reputation and expert attribution matter deeply; authors and organizations face liability for inaccurate or misleading findings. Peer review and institutional review processes also implicitly require human accountability. |
| Adoption barriers | claude-sonnet-5 | 2/5 | No formal licensing requirement exists for writing such reports, though organizational credibility, peer review norms, and reputational risk from errors create moderate friction against full automation. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 3/5 | AI inference costs are minimal, but the integration (prompt engineering, fact-checking, domain adaptation) and required human oversight (verification of claims, data accuracy, policy implications) approach the cost of having a skilled analyst draft sections themselves. |
| Cost vs. human wage | claude-sonnet-5 | 3/5 | AI drafting tools are cheap per word, but the need for expert review, fact-checking, and data verification by a trained analyst keeps overall costs closer to comparable rather than order-of-magnitude cheaper. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 3/5 | LLMs can produce coherent paper drafts and summaries of research, and some organizations use them for initial composition; however, they hallucinate citations, misinterpret statistical results, and lack deep domain verification, making them useful as assistants rather than reliable end-to-end performers in production academic workflows. |
| Technical feasibility today | claude-sonnet-5 | 3/5 | LLM-based writing assistants are widely deployed for drafting technical and academic text, but for specialized climate policy analysis they still require heavy human editing for accuracy, citation integrity, and nuanced interpretation. |
Develop, or contribute to the development of, educational or outreach programs on the environment or climate change.
44CI 39–50 · exposure 30 · augmentation 75 · importance 3.7/5 · click for rater detail
Develop, or contribute to the development of, educational or outreach programs on the environment or climate change.
44| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 3/5 | Environmental and policy organizations are beginning to experiment with AI-assisted content creation and program design, but adoption remains exploratory; full replacement of human program developers is limited by trust, institutional norms, and need for subject-matter expertise. |
| Sector adoption velocity | claude-sonnet-5 | 2/5 | Environmental policy and NGO sectors are slower AI adopters compared to finance or tech, with pilots emerging but production-level integration into program development still rare. |
| Augmentation potential | claude-haiku-4-5-20251001 | 4/5 | AI can significantly amplify a policy analyst's productivity by generating draft curricula, suggesting outreach channels, identifying audience segments, and producing multiple messaging variants, allowing the human to focus on validation and strategic refinement. |
| Augmentation potential | claude-sonnet-5 | 4/5 | AI substantially aids drafting outreach copy, summarizing research, and generating educational content ideas, meaningfully speeding up the analyst's work while they retain strategic control. |
| Task automatability | claude-haiku-4-5-20251001 | 2/5 | AI can draft educational content and generate outreach materials, but the task requires strategic design, stakeholder engagement, and judgment about messaging effectiveness that cannot be fully automated. Significant human oversight and iterative refinement are needed. |
| Task automatability | claude-sonnet-5 | 2/5 | AI can draft outreach materials or lesson plans, but designing an effective program requires stakeholder engagement, contextual judgment, and iterative human decision-making that current AI cannot fully replace end-to-end.wat |
| Adoption barriers | claude-haiku-4-5-20251001 | 2/5 | Educational and outreach programs typically require human judgment and credibility but no strict legal mandate that a human produce them. Organizational preference for human-designed programs and reputational risk provide moderate friction, but regulatory barriers are low. |
| Adoption barriers | claude-sonnet-5 | 2/5 | No licensing requirement mandates a human for this task, though organizational credibility and audience trust in program authorship create some friction against pure automation. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 4/5 | AI-generated drafts and content frameworks cost pennies per instance compared to professional educator salary ($60–90k loaded), making AI dramatically cheaper for the raw material production component, though human oversight remains necessary. |
| Cost vs. human wage | claude-sonnet-5 | 3/5 | AI can cheaply generate drafts of educational materials, but the human oversight, stakeholder coordination, and program design work still dominate the cost, keeping overall savings moderate. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 3/5 | AI systems can generate educational copy, create outlines, and suggest communication strategies, and some organizations use these tools in workflow; however, no mature product reliably produces end-to-end validated educational programs without substantial human review and testing. |
| Technical feasibility today | claude-sonnet-5 | 2/5 | Generative AI tools are used ad hoc to draft content, but no deployed product manages full program development including community engagement and strategic design reliably in production. |
Prepare grant applications to obtain funding for programs related to climate change, environmental management, or sustainability.
42CI 30–55 · exposure 42 · augmentation 75 · importance 3.5/5 · click for rater detail
Prepare grant applications to obtain funding for programs related to climate change, environmental management, or sustainability.
42| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 2/5 | Climate/sustainability organizations are moderate in digital maturity and AI adoption. Grant writing remains manual and human-centric in most NGOs and government agencies; while early adopters experiment with AI drafting assistants, production-level automation is rare. Sectors like academic research and nonprofits lag compared to finance or tech. |
| Sector adoption velocity | claude-sonnet-5 | 3/5 | Nonprofit, environmental, and policy sectors show growing but uneven AI adoption for writing tasks, with many organizations still cautious about full reliance on AI-generated content for funding applications. |
| Augmentation potential | claude-haiku-4-5-20251001 | 4/5 | AI substantially assists human grant writers by drafting sections, summarizing literature, checking budget logic, and formatting—raising productivity markedly while the analyst retains control, judgment, and legal accountability. This is a strong augmentation scenario where human expertise is enhanced rather than replaced. |
| Augmentation potential | claude-sonnet-5 | 4/5 | AI substantially assists drafting narratives, formatting, summarizing data, and adapting boilerplate content to different funders, meaningfully speeding up the analyst's workflow while they retain final control. |
| Task automatability | claude-haiku-4-5-20251001 | 2/5 | Writing grant applications requires significant human judgment about funding priorities, organizational fit, and narrative persuasion. While AI can draft sections (budget justification, literature synthesis), the strategic positioning, stakeholder alignment, and final review demand human expertise. AI cannot reliably handle the full end-to-end task at 50% time savings with equal quality. |
| Task automatability | claude-sonnet-5 | 3/5 | AI can draft substantial portions of grant narratives, budgets, and boilerplate sections given inputs, but tailoring to specific funder priorities, strategic framing, and final judgment calls still require significant human effort. |
| Adoption barriers | claude-haiku-4-5-20251001 | 4/5 | Funders (government agencies, private foundations) typically require explicit sign-off by qualified human staff, organizational leadership, and often legal review. Accountability for grant misrepresentation and fraud liability rest on humans, and institutional policies generally prohibit fully autonomous AI submission. These create strong regulatory and organizational friction. |
| Adoption barriers | claude-sonnet-5 | 2/5 | No licensing requirement for grant writing itself, though funders often expect organizational representation and accountability tied to a named human applicant, creating some institutional friction. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 2/5 | While AI tools (API credits, subscriptions) cost far less per hour than grant writer labor, the loaded cost of human oversight, quality checking, and strategic revision is substantial. The net cost advantage is modest because significant human time remains irreplaceable; the ratio is closer to parity than 10:1. |
| Cost vs. human wage | claude-sonnet-5 | 3/5 | AI can cut drafting time substantially, but the overall grant application process (research, stakeholder input, review, compliance checks) still requires paid analyst time, keeping costs roughly comparable to partial automation savings. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 3/5 | AI writing assistants and document generation tools exist and are used in grant preparation (e.g., ChatGPT for drafting, Overleaf for formatting), but they produce material errors in budget calculations, compliance details, and grant-specific requirements. Deployed products assist but do not reliably complete applications independently; human review remains essential. |
| Technical feasibility today | claude-sonnet-5 | 3/5 | AI writing tools and LLM-based drafting assistants are widely used in grant writing today, but reliability varies and human review/editing is standard practice rather than fully autonomous submission. |
Research policies, practices, or procedures for climate or environmental management.
42CI 25–59 · exposure 38 · augmentation 88 · importance 4.0/5 · click for rater detail
Research policies, practices, or procedures for climate or environmental management.
42| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 2/5 | Government and NGO sectors (where policy analysts concentrate) adopt AI more slowly than tech or finance. Adoption remains primarily pilot-stage for analytical tasks, with limited production-scale displacement visible in public data. |
| Sector adoption velocity | claude-sonnet-5 | 3/5 | Environmental and policy analysis fields are professional-services-adjacent and increasingly using AI tools for research, but adoption lags top digital sectors like finance or tech due to specialized regulatory content needs. |
| Augmentation potential | claude-haiku-4-5-20251001 | 4/5 | AI significantly augments policy analysts by rapidly synthesizing large policy documents, identifying precedents, and surfacing relevant data, allowing humans to focus on critical evaluation and stakeholder consultation. Humans remain essential to judgment; AI amplifies their research capacity. |
| Augmentation potential | claude-sonnet-5 | 5/5 | AI substantially accelerates the literature and policy scanning, summarization, and comparison process, letting analysts focus on judgment and synthesis while remaining in control of conclusions. |
| Task automatability | claude-haiku-4-5-20251001 | 2/5 | AI can assist with literature review, data gathering, and policy document synthesis, but the task requires nuanced evaluation of policy trade-offs, stakeholder impact analysis, and contextual judgment that demands human expertise. Automation would save time on information gathering only—not the core analytical work. |
| Task automatability | claude-sonnet-5 | 3/5 | AI can rapidly gather, summarize, and synthesize policy documents, regulations, and precedents, saving significant research time, but validating accuracy, currency, and jurisdictional nuance still requires human judgment.' |
| Adoption barriers | claude-haiku-4-5-20251001 | 4/5 | Policy analysis often feeds into regulatory decisions and organizational strategy where institutional accountability and professional judgment are required. Liability, regulatory reliance on expert analysis, and organizational preference for credentialed analysts create meaningful adoption friction. |
| Adoption barriers | claude-sonnet-5 | 2/5 | There is no licensing requirement to conduct this research, though organizations may have quality/liability concerns about relying on unverified AI outputs for policy recommendations. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 2/5 | AI-assisted research tools cost less than pure human labor per document processed, but policy analysts command high hourly rates for their specialized expertise. The per-task cost remains comparable when accounting for oversight and validation of AI-generated analysis. |
| Cost vs. human wage | claude-sonnet-5 | 4/5 | AI-assisted research tools dramatically cut the hours needed for literature and policy review compared to a human analyst's fully loaded time, though oversight and fact-checking add some cost back. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 2/5 | While AI tools can summarize policy documents and extract data, deployed systems lack the domain expertise and contextual reasoning needed to independently evaluate policy effectiveness or identify gaps. Existing products serve as research aids rather than reliable end-to-end policy analysis systems. |
| Technical feasibility today | claude-sonnet-5 | 3/5 | General-purpose LLM tools and research assistants are deployed for policy research today, but they have known error rates (hallucinated citations, outdated info) requiring verification, so they are not yet fully reliable standalone products for this niche. |
Prepare study reports, memoranda, briefs, testimonies, or other written materials to inform government or environmental groups on environmental issues, such as climate change.
39CI 25–54 · exposure 38 · augmentation 88 · importance 4.3/5 · click for rater detail
Prepare study reports, memoranda, briefs, testimonies, or other written materials to inform government or environmental groups on environmental issues, such as climate change.
39| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 2/5 | Government agencies and policy organizations have adopted LLM drafting tools for preliminary work, but adoption remains at the pilot and supplementary stage. Deep, production-level replacement is rare because policy outputs require human credibility and accountability. |
| Sector adoption velocity | claude-sonnet-5 | 3/5 | Policy analysis and consulting-adjacent professional services show moderate AI tool adoption for drafting and research, but government and environmental policy work adopts more cautiously due to accountability concerns. |
| Augmentation potential | claude-haiku-4-5-20251001 | 4/5 | AI significantly augments policy analysts by rapidly drafting sections, summarizing research, organizing evidence, and suggesting arguments, substantially accelerating report production while analysts focus on validation, synthesis, and strategic framing. This is one of the stronger augmentation use cases in the policy domain. |
| Augmentation potential | claude-sonnet-5 | 5/5 | AI writing tools substantially speed up drafting, summarizing research, and structuring reports and testimony, while the analyst retains responsibility for judgment, accuracy, and final content. |
| Task automatability | claude-haiku-4-5-20251001 | 2/5 | Large language models can draft sections of policy briefs and memoranda from source materials, but synthesizing multi-source analysis, ensuring factual accuracy on complex climate data, and meeting specific stakeholder requirements demands substantial human oversight and revision. No current system achieves ≥50% time savings at equal quality end-to-end for government-quality policy documents. |
| Task automatability | claude-sonnet-5 | 3/5 | AI can draft substantial portions of reports and briefs from source material, but synthesizing policy-relevant judgment, verifying accuracy, and tailoring to specific political/legal contexts still requires significant human oversight, limiting full end-to-end automation. |
| Adoption barriers | claude-haiku-4-5-20251001 | 4/5 | Government and environmental organizations require accountability, official sign-off, and legal responsibility for testimony and formal policy advice; a licensed analyst or senior official must ultimately author and vouch for the content. Liability and regulatory requirements for accuracy on climate science create hard barriers to full automation. |
| Adoption barriers | claude-sonnet-5 | 3/5 | No formal licensing is required to write these documents, but institutional review, accountability for accuracy in government submissions, and reputational/legal risk create moderate friction against pure AI authorship. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 2/5 | AI can reduce drafting labor costs modestly, but the need for expert policy analysts to verify claims, integrate nuanced data, and ensure accuracy means total cost (inference + integration + expert review) remains high relative to direct wage savings. Integration and oversight costs are substantial. |
| Cost vs. human wage | claude-sonnet-5 | 4/5 | Drafting assistance via AI tools is far cheaper per word than analyst time, though the need for expert review and fact-checking narrows the effective savings compared to a fully hands-off process. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 2/5 | While LLMs can generate draft text and assist with research synthesis, no deployed product reliably produces publication-ready policy reports without significant human editing, fact-checking, and rewriting. Existing tools lack the domain depth and accountability requirements for official government testimony or briefs. |
| Technical feasibility today | claude-sonnet-5 | 3/5 | LLM-based writing assistants are widely deployed for drafting technical and policy documents, but factual reliability, citation accuracy, and nuanced policy framing still require heavy human review, especially for authoritative government materials. |
Analyze and distill climate-related research findings to inform legislators, regulatory agencies, or other stakeholders.
38CI 30–46 · exposure 34 · augmentation 88 · importance 4.3/5 · click for rater detail
Analyze and distill climate-related research findings to inform legislators, regulatory agencies, or other stakeholders.
38| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 2/5 | Policy and regulatory organizations remain conservative in automation adoption; climate analysis directly informs high-stakes decisions, so human analyst involvement is still the norm despite pilot projects; laggard digitization in government limits rapid deployment. |
| Sector adoption velocity | claude-sonnet-5 | 3/5 | Policy and research-adjacent professional services are moderately fast adopters of AI drafting tools, though government and regulatory environments tend to lag private-sector adoption rates. |
| Augmentation potential | claude-haiku-4-5-20251001 | 4/5 | AI excels at helping analysts quickly identify relevant research, generate initial drafts of distillations, and flag key findings from large literature sets, substantially boosting an analyst's capacity to cover more research and produce faster briefings while they retain judgment on policy implications. |
| Augmentation potential | claude-sonnet-5 | 5/5 | AI substantially accelerates literature review, summarization, and first-draft synthesis, letting analysts focus on judgment, framing, and stakeholder communication while remaining in the loop. |
| Task automatability | claude-haiku-4-5-20251001 | 2/5 | While AI can summarize and synthesize research findings effectively, the task requires judgment about which findings matter for specific policy contexts, stakeholder priorities, and legislative feasibility—nuanced decisions that current systems cannot reliably make end-to-end without substantial human oversight and revision. |
| Task automatability | claude-sonnet-5 | 2/5 | AI can summarize and synthesize research documents, but the analytical judgment, contextualization to policy needs, and stakeholder-specific framing required limits full automation to well under the 50% end-to-end bar without significant human oversight. |
| Adoption barriers | claude-haiku-4-5-20251001 | 4/5 | Legislators and regulatory agencies require credible, accountable analysis that can withstand legal and public scrutiny; policy decisions tied to AI-distilled findings create liability and reputational risk, and stakeholders often demand attribution to named experts rather than AI summaries. |
| Adoption barriers | claude-sonnet-5 | 2/5 | No formal licensing requirement exists for this analytical work, though reputational and institutional trust factors create some friction against fully autonomous AI-generated policy briefs. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 2/5 | AI inference is cheap, but integration into policy workflows, fact-checking outputs against primary research, and human oversight to ensure stakeholder-appropriate framing add significant cost, making the all-in solution only modestly cheaper than analyst time. |
| Cost vs. human wage | claude-sonnet-5 | 3/5 | AI can cheaply draft summaries but human expert review, fact-checking, and stakeholder tailoring add substantial cost, making the all-in cost roughly comparable to a skilled analyst for high-stakes outputs. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 3/5 | AI tools (LLMs, summarization systems) can distill research content and produce draft briefings, but deployed products lack the contextual understanding to ensure accuracy on complex climate science and appropriateness for diverse regulatory audiences; human verification remains necessary. |
| Technical feasibility today | claude-sonnet-5 | 3/5 | LLM-based tools are widely deployed for literature synthesis and summarization, but reliably distilling nuanced climate science into policy-relevant insights with accuracy still requires human validation, especially given risk of hallucination on technical data. |
Make legislative recommendations related to climate change or environmental management, based on climate change policies, principles, programs, practices, and processes.
27CI 25–29 · exposure 25 · augmentation 75 · importance 4.3/5 · click for rater detail
Make legislative recommendations related to climate change or environmental management, based on climate change policies, principles, programs, practices, and processes.
27| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 2/5 | Policy organizations are cautious adopters of AI for core recommendation work; while data analysis and research support are accelerating, end-to-end automation of legislative recommendation-making remains rare, reflecting both technical and institutional hesitation. |
| Sector adoption velocity | claude-sonnet-5 | 2/5 | Government and policy sectors are historically slow adopters of AI tools compared to finance or tech, with pilots emerging but production use in legislative drafting still limited. |
| Augmentation potential | claude-haiku-4-5-20251001 | 4/5 | AI significantly augments policy analysts by rapidly synthesizing research, comparing policy frameworks, and generating option analysis, allowing analysts to focus on interpretation, stakeholder engagement, and value-informed tradeoffs—a strong productivity multiplier with humans retaining decision authority. |
| Augmentation potential | claude-sonnet-5 | 4/5 | AI can meaningfully assist analysts by summarizing scientific literature, comparing policy frameworks, and drafting initial recommendation language, significantly speeding up research and writing phases. |
| Task automatability | claude-haiku-4-5-20251001 | 2/5 | While AI can synthesize climate policy research and generate policy summaries, making substantive legislative recommendations requires weighing competing values, political feasibility, and stakeholder impacts—tasks demanding human judgment and accountability that current AI cannot reliably perform end-to-end at the quality threshold needed for policy contexts. |
| Task automatability | claude-sonnet-5 | 2/5 | Drafting legislative language and synthesizing input can be AI-assisted, but final recommendations require political judgment, stakeholder weighing, and accountability that current AI cannot reliably replicate end-to-end. |
| Adoption barriers | claude-haiku-4-5-20251001 | 4/5 | Legislative recommendations carry legal, fiduciary, and reputational liability; organizations and elected officials require a human expert to take responsibility and sign off, and regulatory or professional norms expect human judgment on matters of public policy significance. |
| Adoption barriers | claude-sonnet-5 | 4/5 | Legislative recommendations often require credentialed policy expertise, agency sign-off, and accountability to elected officials or the public, creating strong institutional and legal barriers to full automation. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 2/5 | Policy analysis typically involves senior specialists earning high salaries; while AI research assistance can reduce labor on components, the full task—justifiable recommendations tied to political and legal risk—still requires expert oversight that keeps total cost high relative to AI inference savings. |
| Cost vs. human wage | claude-sonnet-5 | 3/5 | AI can cheaply generate drafts and research summaries, but the cost of expert review, legal vetting, and political vetting remains substantial, keeping overall cost roughly comparable to human-led work. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 2/5 | AI systems can assist with research synthesis and draft policy options, but no deployed product independently makes legislative recommendations that organizations rely on as authoritative policy counsel; such work remains human-led with AI in a support role only. |
| Technical feasibility today | claude-sonnet-5 | 2/5 | Products exist that can summarize policy documents and draft memos, but no deployed system autonomously produces vetted legislative recommendations used in real policymaking without heavy human revision. |
Promote initiatives to mitigate climate change with government or environmental groups.
18CI 11–25 · exposure 8 · augmentation 63 · importance 4.1/5 · click for rater detail
Promote initiatives to mitigate climate change with government or environmental groups.
18| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 2/5 | Climate policy organizations are slowly piloting AI for data analysis and drafting support, but actual adoption of AI for initiative promotion itself remains minimal; this remains a human-centered sector with limited automation in production. |
| Sector adoption velocity | claude-sonnet-5 | 2/5 | Policy and advocacy work in government/NGO sectors adopts AI mainly for drafting and research support; direct advocacy automation is rare and slow to spread. |
| Augmentation potential | claude-haiku-4-5-20251001 | 3/5 | AI can meaningfully assist by generating policy summaries, analyzing climate datasets, identifying stakeholder networks, and drafting outreach materials, thereby raising human analyst productivity without replacing the essential human promotion role. |
| Augmentation potential | claude-sonnet-5 | 4/5 | AI can help draft talking points, policy briefs, persuasive materials, and analyze stakeholder positions, meaningfully boosting the analyst's efficiency in preparing for advocacy. |
| Task automatability | claude-haiku-4-5-20251001 | 2/5 | While AI can draft policy briefs, synthesize climate data, and model scenarios, promoting initiatives requires sustained stakeholder engagement, political negotiation, and building coalitions—activities demanding human judgment, relationship-building, and accountability that current AI cannot perform end-to-end with 50% time savings at equal quality. |
| Task automatability | claude-sonnet-5 | 1/5 | Promoting initiatives requires relationship-building, negotiation, public speaking, and persuasion in live political/organizational contexts that AI cannot perform end-to-end today. |
| Adoption barriers | claude-haiku-4-5-20251001 | 4/5 | Promotion of climate initiatives often requires human credibility, institutional authority, and legal accountability; many government and nonprofit roles mandate a qualified human representative as the face and voice of advocacy efforts. |
| Adoption barriers | claude-sonnet-5 | 3/5 | No licensing requirement per se, but effective advocacy depends on human trust, credibility, and relationships that create strong organizational and social barriers to substitution. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 2/5 | AI can reduce research and drafting costs (raising the ratio somewhat), but the core task of promotion—securing buy-in from decision-makers—remains human-dependent, so all-in costs remain comparable to or exceed human labor for the full task. |
| Cost vs. human wage | claude-sonnet-5 | 2/5 | AI can cheaply generate supporting materials, but the core advocacy work still requires paid human time, so overall cost savings are limited. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 1/5 | No deployed product autonomously promotes policy initiatives with government or environmental groups; this task fundamentally involves human persuasion, advocacy, and institutional relationships that fall outside current AI capabilities in production systems. |
| Technical feasibility today | claude-sonnet-5 | 1/5 | No deployed product independently advocates for policy initiatives with government or environmental groups; this remains a human relational and advocacy function. |
Present climate-related information at public interest, governmental, or other meetings.
16CI 7–25 · exposure 13 · augmentation 75 · importance 4.2/5 · click for rater detail
Present climate-related information at public interest, governmental, or other meetings.
16| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 2/5 | Government and policy sectors adopt automation slowly for public-facing roles; most climate policy presentations remain human-delivered. While AI drafting tools are emerging in policy offices, autonomous presentation by AI at meetings is not yet a measurable trend in production deployments. |
| Sector adoption velocity | claude-sonnet-5 | 2/5 | Government and public-facing policy meetings are slow-moving, relationship-driven contexts with limited AI agent deployment for direct representation. |
| Augmentation potential | claude-haiku-4-5-20251001 | 4/5 | AI can significantly assist by generating data visualizations, drafting talking points, summarizing climate datasets, and preparing responses to anticipated questions, substantially raising a policy analyst's preparation efficiency and presentation quality while the human retains full control and delivery responsibility. |
| Augmentation potential | claude-sonnet-5 | 4/5 | AI can strongly assist in drafting slides, summarizing data, anticipating questions, and tailoring messaging, meaningfully boosting the analyst's efficiency and preparation quality. |
| Task automatability | claude-haiku-4-5-20251001 | 2/5 | While AI can generate climate data summaries and draft presentation materials, the task fundamentally requires a human expert to deliver information, respond to live questions, and navigate the political and stakeholder dynamics of public meetings. AI cannot meaningfully replace the presenter's credibility, judgment, and real-time interaction. |
| Task automatability | claude-sonnet-5 | 1/5 | Live public presentation and real-time engagement with governmental audiences requires human presence, credibility, and adaptive interpersonal skills that current AI cannot replicate end-to-end. |
| Adoption barriers | claude-haiku-4-5-20251001 | 4/5 | Strong barriers exist: public and governmental meetings expect a credentialed human expert to present, there is reputational and liability risk if AI-generated content proves inaccurate, and stakeholder trust typically requires human accountability and real-time responsiveness that regulation and institutional norms reinforce. |
| Adoption barriers | claude-sonnet-5 | 4/5 | Public accountability, trust, and often statutory requirements for a named human representative to present and answer questions create strong barriers to substitution. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 2/5 | The cost of AI-generated presentation materials plus human presenter time is comparable to or exceeds the cost of a human analyst preparing and delivering the presentation, since the human remains essential and oversight of AI outputs adds overhead. |
| Cost vs. human wage | claude-sonnet-5 | 1/5 | AI cannot substitute for the human presenter, so the relevant cost is the human's full loaded wage plus any minimal AI tool cost for prep, not a replacement cost comparison. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 2/5 | No deployed product reliably performs end-to-end public presentation of climate policy at scale. AI can assist with slide generation and data visualization, but production systems do not yet autonomously present at government or public interest meetings with acceptable credibility and adaptability. |
| Technical feasibility today | claude-sonnet-5 | 1/5 | No deployed product autonomously represents an organization at public or government meetings; this remains firmly a human role. |
Present and defend proposals for climate change research projects.
12CI 7–16 · exposure 5 · augmentation 75 · importance 3.6/5 · click for rater detail
Present and defend proposals for climate change research projects.
12| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 2/5 | Climate policy organizations and research institutions still expect human experts to present and defend proposals in real settings (public hearings, review panels, funding agencies). Adoption of AI-only presentation is minimal; AI is used at most as a drafting assistant, not a replacement for the human defender. |
| Sector adoption velocity | claude-sonnet-5 | 2/5 | Policy analysis and research funding sectors are adopting AI for drafting and research support, but live proposal defense remains a low-adoption area with minimal displacement evidence. |
| Augmentation potential | claude-haiku-4-5-20251001 | 4/5 | AI can meaningfully assist by generating draft proposals, summarizing rebuttals, synthesizing research findings, and suggesting argument structures—all of which can significantly boost a human analyst's preparation and presentation effectiveness. The human remains the credible defender, but AI can substantially amplify their productivity. |
| Augmentation potential | claude-sonnet-5 | 4/5 | AI can strongly assist in drafting proposal content, anticipating objections, generating supporting data/visuals, and rehearsing responses, meaningfully boosting the analyst's preparation and effectiveness. |
| Task automatability | claude-haiku-4-5-20251001 | 1/5 | Presenting and defending research proposals requires synthesizing complex arguments, anticipating counterarguments, reading audience response, and adjusting rhetoric in real time—capabilities that current AI systems lack. While AI can draft proposal text, the live defense and persuasive adaptation demanded here remains a fundamentally human communicative task. |
| Task automatability | claude-sonnet-5 | 1/5 | Presenting and defending proposals requires live persuasion, real-time rebuttal, and credibility with an audience—capabilities current AI cannot perform end-to-end in place of a human presenter. |
| Adoption barriers | claude-haiku-4-5-20251001 | 4/5 | Presenting and defending research proposals typically requires the analyst to be a credentialed expert accountable for claims made, and organizations place high trust and liability value on having a human expert visibly stand behind a proposal. Regulatory and reputational pressure strongly favors human authorship and presentation. |
| Adoption barriers | claude-sonnet-5 | 4/5 | Defending research proposals often requires accountable, credentialed experts who can be questioned and held responsible for claims and funding decisions, creating strong organizational and credibility barriers to AI substitution. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 2/5 | AI can assist with drafting at low cost, but full replacement would require a human to oversee and validate the defense anyway, and human policy analysts remain much cheaper to deploy for the actual presentation-and-defense act itself. The overhead of AI oversight doesn't yet beat direct human labor. |
| Cost vs. human wage | claude-sonnet-5 | 1/5 | Since AI cannot substitute for the human presenter/defender role, there is no viable cost comparison—the human must still perform the task, making AI substitution costs irrelevant or additive. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 2/5 | No deployed product reliably presents and defends proposals end-to-end. AI can generate draft text or talking points, but defending requires authentic engagement with a live audience, handling hostile or nuanced questions, and demonstrating genuine expertise conviction—elements current systems cannot authentically execute at production scale. |
| Technical feasibility today | claude-sonnet-5 | 1/5 | No deployed product autonomously presents and defends research proposals to stakeholders or funding bodies; this remains a human-led interpersonal activity. |
Related occupations — Life, Physical & Social Science
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.