Economists
19-3011.00Median wage $124,720/yr17,790 employed (US)Rank #250 of 923 scored · top 27% by substitution
Conduct research, prepare reports, or formulate plans to address economic problems related to the production and distribution of goods and services or monetary and fiscal policy. May collect and process economic and statistical data using sampling techniques and econometric methods.
Sub-scores
0–100 · band = confidence interval from rater disagreement
Substitution36
Exposure33
Augmentation78
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%32
panel mean rating 2.3/5 → substitution pressure 32/100
Cost vs. human wagew 15%36
panel mean rating 2.5/5 → substitution pressure 36/100
Adoption barriersw 20%inverted — strong barriers lower the score42
panel mean rating 3.3/5 (barrier strength) → substitution pressure 42/100
Sector adoption velocityw 10%42
panel mean rating 2.7/5 → substitution pressure 42/100
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.
Review documents written by others.
77CI 76–79 · exposure 75 · augmentation 100 · importance 4.0/5 · click for rater detail
Review documents written by others.
77| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 3/5 | Economics departments and think tanks are beginning to pilot AI-assisted document screening, but production adoption remains piecemeal. Finance and policy firms move faster; academia lags. Overall trajectory is middling, with pilots common but wholesale displacement still rare. |
| Sector adoption velocity | claude-sonnet-5 | 4/5 | Economists work in finance, consulting, government, and research—sectors with above-average AI tool adoption for drafting and reviewing text-heavy outputs. |
| Augmentation potential | claude-haiku-4-5-20251001 | 5/5 | AI document reviewers powerfully augment economists by rapidly extracting summaries, highlighting contradictions, and flagging relevant citations, freeing humans to focus on critical appraisal and synthesis. This is among the highest-impact augmentation use cases in the profession. |
| Augmentation potential | claude-sonnet-5 | 5/5 | AI is highly effective as a first-pass reviewer, flagging errors, inconsistencies, and clarity issues, substantially speeding up the human review process while the economist retains final judgment. |
| Task automatability | claude-haiku-4-5-20251001 | 4/5 | Current AI systems can reliably summarize, extract key claims, identify logical inconsistencies, and flag methodological issues in documents at substantial time savings. However, nuanced judgment about novelty, significance, and deep theoretical critique typically requires human economist oversight, preventing full end-to-end automation. |
| Task automatability | claude-sonnet-5 | 4/5 | Reviewing documents (reports, papers, memos) for clarity, consistency, factual and logical errors is well within current LLM capabilities, especially for grammar, structure, and surface-level content checks, though deep economic reasoning review still benefits from human oversight. |
| Adoption barriers | claude-haiku-4-5-20251001 | 2/5 | Document review is a judgment task often delegated to junior economists; no legal requirement mandates a credentialed economist perform it. Organizational friction (preference for human review in peer-review or compliance contexts) exists but is not a hard barrier. |
| Adoption barriers | claude-sonnet-5 | 2/5 | No licensing requirement mandates human-only review of most economic documents; some organizational policies may require senior economist sign-off, but this is soft friction rather than a hard legal barrier. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 5/5 | API costs for reviewing a document (even lengthy ones) run cents to low dollars, compared to an economist's hourly wage (typically $50–150+ loaded), yielding cost ratios of 1:100 or better per equivalent review pass. |
| Cost vs. human wage | claude-sonnet-5 | 5/5 | AI document review costs pennies per document versus an economist's hourly loaded wage, making it dramatically cheaper for first-pass or supplementary review. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 4/5 | Deployed LLM-based document review tools (including Claude, GPT-4, and specialized legal/research platforms) demonstrably handle document summarization, classification, and initial quality assessment in production. Error rates on factual extraction are low, though subtle economic interpretation remains variable. |
| Technical feasibility today | claude-sonnet-5 | 4/5 | Products like ChatGPT, Claude, and specialized writing-assistant tools are already deployed widely for document review, summarization, and editing feedback in professional settings, though nuanced economic critique still has error rates. |
Study economic and statistical data in area of specialization, such as finance, labor, or agriculture.
68CI 57–79 · exposure 62 · augmentation 100 · importance 4.6/5 · click for rater detail
Study economic and statistical data in area of specialization, such as finance, labor, or agriculture.
68| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 4/5 | Academic and financial institutions, government statistical agencies, and policy organizations are rapidly adopting AI-assisted data analysis tools and platforms; adoption is deepening in digitized sectors where economists work. |
| Sector adoption velocity | claude-sonnet-5 | 4/5 | Finance, economics, and research-heavy sectors have been comparatively fast adopters of AI-driven data analysis and forecasting tools, reflecting the broader professional-services adoption pattern. |
| Augmentation potential | claude-haiku-4-5-20251001 | 5/5 | AI dramatically augments economists' productivity by automating data retrieval, exploratory analysis, and hypothesis testing, allowing economists to focus on interpretation, causal reasoning, and policy implications while remaining firmly in the loop. |
| Augmentation potential | claude-sonnet-5 | 5/5 | AI dramatically accelerates data cleaning, statistical modeling, and literature synthesis, letting economists focus on interpretation and specialized judgment while staying fully in the loop. |
| Task automatability | claude-haiku-4-5-20251001 | 4/5 | AI systems can automatically retrieve, clean, and analyze large datasets, perform statistical tests, generate visualizations, and summarize findings with significant time savings. However, the task requires domain expertise and judgment in interpreting results within broader economic theory, which limits it from a full 5. |
| Task automatability | claude-sonnet-5 | 3/5 | AI can rapidly process, summarize, and identify patterns in economic/statistical datasets, but genuine domain specialization and judgment about which relationships matter still require human oversight, so only part of the workflow meets the 50% time-saving bar. |
| Adoption barriers | claude-haiku-4-5-20251001 | 2/5 | Few regulatory or legal barriers prevent automation of data analysis itself. However, professional economists' sign-off on conclusions and peer-review expectations create modest organizational friction and reputational concerns around fully automated outputs. |
| Adoption barriers | claude-sonnet-5 | 2/5 | No licensing requirement restricts who can study economic data, though some organizational or reputational preference for credentialed economists exists, especially in policy-sensitive contexts. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 5/5 | AI-powered data analysis via cloud services or open-source tools costs orders of magnitude less than hiring an economist for routine data processing, cleaning, visualization, and basic statistical work. |
| Cost vs. human wage | claude-sonnet-5 | 3/5 | AI tools reduce time spent on data wrangling and exploratory analysis substantially, but the need for economist oversight, model validation, and interpretation keeps overall costs only moderately lower than fully human-driven analysis. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 4/5 | Deployed products (statistical software with AI assistance, data analytics platforms, and LLMs augmented with data tools) reliably perform data analysis, querying, and initial statistical modeling at scale. Some gaps remain in automated contextual interpretation and handling novel data structures, preventing a 5. |
| Technical feasibility today | claude-sonnet-5 | 3/5 | Products like Copilot/ChatGPT with code interpreter, and specialized tools (e.g., statistical/econometric assistants), can perform data analysis today, but reliable end-to-end specialized economic research still shows material error rates and requires expert validation. |
Compile, analyze, and report data to explain economic phenomena and forecast market trends, applying mathematical models and statistical techniques.
66CI 57–75 · exposure 62 · augmentation 100 · importance 4.3/5 · click for rater detail
Compile, analyze, and report data to explain economic phenomena and forecast market trends, applying mathematical models and statistical techniques.
66| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 4/5 | Financial services, consulting, and quantitative research firms are rapidly deploying AI for data analysis and forecasting; adoption is production-grade in information-sector firms. Government and academic sectors lag but are accelerating adoption of AI-assisted modeling. |
| Sector adoption velocity | claude-sonnet-5 | 4/5 | Economics sits within finance and professional/research services, sectors with fast, deep AI tool adoption for data analysis, forecasting, and reporting, though full end-to-end deployment remains less common than augmentation. |
| Augmentation potential | claude-haiku-4-5-20251001 | 5/5 | AI substantially amplifies economist productivity by automating data wrangling, hypothesis testing, and draft forecasts, freeing them to focus on causal reasoning, interpretation, and strategic insights. This is a canonical augmentation case where AI excels at routine analytics while economists retain judgment. |
| Augmentation potential | claude-sonnet-5 | 5/5 | AI substantially enhances economists' productivity by automating data cleaning, running statistical models, generating visualizations, and drafting narrative explanations, while the economist retains interpretive and judgment control. |
| Task automatability | claude-haiku-4-5-20251001 | 4/5 | Current AI can compile structured economic data, run standard statistical analyses, fit mathematical models, and generate forecasts at scale with >50% time savings compared to manual analysis. However, model selection, assumption validation, and interpretation of novel or anomalous data typically require human oversight, preventing a full end-to-end 5. |
| Task automatability | claude-sonnet-5 | 3/5 | AI can automate data compilation, statistical modeling, and drafting of explanatory reports, but selecting appropriate models, interpreting ambiguous economic phenomena, and validating forecasts against domain judgment still require significant human oversight and setup. |
| Adoption barriers | claude-haiku-4-5-20251001 | 2/5 | Few legal barriers exist to automating this task; however, client trust, regulatory appetite for algorithmic forecasts, and organizational reliance on economist credibility create moderate friction. Central banks and policy institutions often require economist sign-off, limiting pure displacement. |
| Adoption barriers | claude-sonnet-5 | 2/5 | There's no licensing requirement to perform economic analysis, but organizational reliance on credentialed expertise and reputational/liability concerns around forecast accuracy create moderate friction against pure automation. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 4/5 | AI inference, model fitting, and report generation cost far less than hiring economists for data processing and standard forecasting tasks. At typical economist wages ($70k–$150k+ fully loaded), cloud-based analytics easily achieve 5–10× cost advantage for routine work, though bespoke research may narrow the gap. |
| Cost vs. human wage | claude-sonnet-5 | 3/5 | AI can cut costs on data processing and drafting substantially, but integration with proprietary datasets, model validation, and expert interpretation still require costly human economist time, making the overall ratio moderate rather than dramatically cheaper. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 4/5 | Mature products (Python/R stacks, specialized econometric platforms, and LLM-assisted analytics) reliably perform data compilation, statistical testing, and forecasting in production across finance and government agencies. Performance is strong on routine analyses, though complex scenario modeling or causal inference still carries material error rates without human review. |
| Technical feasibility today | claude-sonnet-5 | 3/5 | Deployed tools (e.g., AI-assisted statistical software, LLM-based analysis platforms) can generate forecasts and reports, but production use in economics still shows material error rates in judgment-heavy interpretation and model selection, requiring expert review. |
Forecast production and consumption of renewable resources and supply, consumption, and depletion of non-renewable resources.
42CI 32–51 · exposure 38 · augmentation 75 · importance 2.8/5 · click for rater detail
Forecast production and consumption of renewable resources and supply, consumption, and depletion of non-renewable resources.
42| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 3/5 | Energy, finance, and government agencies pilot ML-based forecasting, but adoption of fully automated resource forecasts remains limited. Most organizations still rely on economist-led processes with AI as a support tool rather than a replacement, reflecting cautious adoption in a high-stakes domain. |
| Sector adoption velocity | claude-sonnet-5 | 3/5 | Economics and research organizations are adopting AI/ML tools for forecasting and data analysis at a moderate pace, with pilots and augmented workflows more common than full automation. |
| Augmentation potential | claude-haiku-4-5-20251001 | 4/5 | AI significantly assists economists by automating data preprocessing, generating baseline forecasts, stress-testing scenarios, and visualizing outcomes. These capabilities meaningfully raise productivity while economists focus on interpretation, assumption-setting, and judgment calls—a strong augmentation profile. |
| Augmentation potential | claude-sonnet-5 | 4/5 | AI substantially assists economists by automating data gathering, running simulations, and drafting scenario narratives, meaningfully speeding up the forecasting workflow while the economist retains final judgment. |
| Task automatability | claude-haiku-4-5-20251001 | 3/5 | AI can automate parts of this task—historical data analysis, trend extrapolation, and scenario modeling using statistical/ML methods—but requires domain expertise to validate assumptions, incorporate policy shifts, and handle non-linear tipping points. End-to-end forecasting without human oversight would miss critical context and produce unreliable predictions. |
| Task automatability | claude-sonnet-5 | 2/5 | AI can assist with data processing and running forecasting models, but the task requires integrating domain judgment, structural economic assumptions, and novel scenario reasoning that current systems cannot fully replace end-to-end at production quality. |
| Adoption barriers | claude-haiku-4-5-20251001 | 3/5 | Policy and investment decisions depend on these forecasts; organizations often require economist credentialing and accountability. However, there are no hard legal barriers preventing AI-assisted or AI-driven forecasting, only institutional preferences for human judgment and liability concerns. |
| Adoption barriers | claude-sonnet-5 | 2/5 | No licensing requirement mandates a human economist sign off on forecasts, but institutional trust, reputational risk, and the need for defensible methodology create moderate organizational friction against pure automation. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 2/5 | AI inference is cheap, but integration into econometric pipelines, data curation, validation, and expert oversight add substantial cost. The loaded wage for an economist remains high, and the residual human effort required keeps total AI-automation cost closer to human cost than substantially below it. |
| Cost vs. human wage | claude-sonnet-5 | 2/5 | Building and validating credible resource forecasting models still requires significant economist time for specification, data curation, and interpretation, so AI reduces but does not eliminate the dominant labor cost. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 3/5 | Products exist for time-series forecasting and resource modeling (e.g., specialized econometric software, energy forecasting platforms), but they require significant expert calibration and typically achieve moderate error rates on novel scenarios. No fully autonomous system reliably produces publication-grade resource forecasts without expert review. |
| Technical feasibility today | claude-sonnet-5 | 2/5 | Some econometric and forecasting tools exist and are used to support analysts, but no deployed product independently produces reliable, defensible resource forecasts without expert oversight and model specification by humans. |
Explain economic impact of policies to the public.
29CI 25–34 · exposure 25 · augmentation 75 · importance 4.0/5 · click for rater detail
Explain economic impact of policies to the public.
29| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 2/5 | Economists and policy institutions remain slow to automate public communication; they value human credibility and are risk-averse about AI-generated public messaging. Pilot projects exist, but production adoption in economics and government sectors lags white-collar automation more broadly. |
| Sector adoption velocity | claude-sonnet-5 | 2/5 | Economic communication and policy sectors (government, academia, think tanks) are moderate-to-slow adopters of AI-driven public communication compared to tech/finance sectors. |
| Augmentation potential | claude-haiku-4-5-20251001 | 4/5 | AI excels at augmenting this task: generating first drafts of policy summaries, analyzing economic data to support narrative, identifying key talking points, and suggesting alternative phrasings. An economist can use these outputs to work faster while maintaining authority and accuracy. |
| Augmentation potential | claude-sonnet-5 | 4/5 | AI is very useful for drafting explanations, simplifying jargon, generating visualizations, and tailoring messages to audiences, while the economist retains final judgment and voice. |
| Task automatability | claude-haiku-4-5-20251001 | 2/5 | While AI can generate policy summaries and statistical outputs, explaining policy impact to the public requires translating technical concepts into accessible language, navigating cultural context, and responding to audience pushback—tasks that demand human judgment and credibility. AI alone cannot reliably achieve the 50% time savings threshold for this fundamentally persuasive, context-dependent task. |
| Task automatability | claude-sonnet-5 | 2/5 | AI can draft explanations of policy impacts, but crafting authoritative public-facing economic communication requires judgment, credibility, and context-specific nuance that current systems cannot fully replace at equal quality. |
| Adoption barriers | claude-haiku-4-5-20251001 | 4/5 | Public-facing policy explanation often carries regulatory or institutional approval requirements (central banks, government agencies, professional standards), and there is strong organizational and stakeholder preference for a credible human expert to own the explanation. Liability for misleading policy communication is asymmetric. |
| Adoption barriers | claude-sonnet-5 | 3/5 | No strict licensing requirement to explain policy, but institutional credibility, reputational risk, and public trust favor named human experts, especially in official contexts. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 2/5 | AI inference for drafting explanations is cheap, but integrating oversight (fact-checking, tone calibration, audience testing) and the reputational cost of errors mean the full-cost comparison remains unfavorable compared to a human economist's loaded wage on this trust-critical task. |
| Cost vs. human wage | claude-sonnet-5 | 3/5 | AI drafting is cheap, but the value of an economist's task lies in credibility and accountability, so oversight and verification costs keep the ratio only moderately favorable. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 2/5 | No production system reliably performs public-facing policy explanation as a standalone service. AI can draft explanatory text or generate talking points, but these typically require substantial human editing for accuracy, tone, and audience-appropriateness. Deployed chatbots lack the domain credibility and accountability economists possess. |
| Technical feasibility today | claude-sonnet-5 | 2/5 | LLMs can generate plausible explanations of economic policy impacts, but no deployed product reliably serves as the authoritative public communicator on economics; accuracy and nuance issues persist. |
Study the socioeconomic impacts of new public policies, such as proposed legislation, taxes, services, and regulations.
29CI 25–32 · exposure 25 · augmentation 63 · importance 4.2/5 · click for rater detail
Study the socioeconomic impacts of new public policies, such as proposed legislation, taxes, services, and regulations.
29| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 2/5 | Adoption remains slow in core policy analysis. While some government and think-tank economists use AI for data wrangling and literature review, the authoritative impact assessment work itself remains human-driven due to accountability requirements and the stakes of policy decisions. |
| Sector adoption velocity | claude-sonnet-5 | 3/5 | Economics and policy research sectors (government, think tanks, consultancies) are adopting AI tools for research assistance at a middling pace, with pilots common but full production deployment still limited. |
| Augmentation potential | claude-haiku-4-5-20251001 | 3/5 | AI augments economists well on literature synthesis, rapid statistical exploratory analysis, and scenario modeling assistance, but does not transform the core task of causal reasoning and policy trade-off evaluation. Useful support on parts, not the whole. |
| Augmentation potential | claude-sonnet-5 | 4/5 | AI substantially aids literature synthesis, data analysis, drafting, and scenario modeling, meaningfully boosting economist productivity while the human retains responsibility for judgment and conclusions. |
| Task automatability | claude-haiku-4-5-20251001 | 2/5 | AI can assist in data gathering, literature review, and preliminary statistical analysis, but evaluating socioeconomic impacts requires contextual judgment, stakeholder understanding, and normative tradeoffs that current systems cannot reliably perform end-to-end. The task demands synthesizing complex causal chains and policy trade-offs that exceed 50% time savings at equal quality. |
| Task automatability | claude-sonnet-5 | 2/5 | AI can assist with literature review, data pulls, and drafting sections, but rigorous socioeconomic impact analysis requires original modeling, causal judgment, and contextual expertise that current systems cannot fully replace end-to-end. |
| Adoption barriers | claude-haiku-4-5-20251001 | 4/5 | Strong barriers exist: policy impact work is often commissioned by government or legislative bodies that legally require credentialed economists to sign off on findings; liability for incorrect policy analysis is high; and institutional trust in human expertise over algorithmic output remains strong in government and regulatory contexts. |
| Adoption barriers | claude-sonnet-5 | 3/5 | No licensing requirement mandates a human economist, but institutional expectations, credibility requirements for policy analysis, and accountability for public decisions create moderate friction against full automation. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 2/5 | AI tools (data processing, modeling support) have low marginal cost, but the core task—expert economic analysis and judgment—still requires senior economist time. Integration, validation, and oversight costs remain substantial relative to incremental AI value in policy impact studies. |
| Cost vs. human wage | claude-sonnet-5 | 2/5 | AI can cut research and drafting time, but the need for expert validation, custom econometric modeling, and domain judgment keeps overall costs closer to human-comparable rather than order-of-magnitude cheaper. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 2/5 | No deployed product performs comprehensive socioeconomic impact analysis end-to-end with the rigor required for policy work. Tools exist for data processing and visualization, but actual impact assessment still requires human economists to frame questions, validate assumptions, and interpret results in production settings. |
| Technical feasibility today | claude-sonnet-5 | 2/5 | Products like ChatGPT/Claude with data analysis plugins can support parts of this work, but no deployed system reliably conducts full policy impact studies without heavy expert oversight. |
Provide advice and consultation on economic relationships to businesses, public and private agencies, and other employers.
29CI 25–32 · exposure 25 · augmentation 75 · importance 4.0/5 · click for rater detail
Provide advice and consultation on economic relationships to businesses, public and private agencies, and other employers.
29| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 2/5 | Consulting and economics firms are piloting AI for report drafting and background analysis, but actual deployment of AI as the primary advisor is minimal; the professional services sector overall moves cautiously on high-stakes advice, with humans still required in the loop for client-facing work. |
| Sector adoption velocity | claude-sonnet-5 | 3/5 | Professional services and consulting firms are adopting AI tools moderately fast for research and drafting, but full consultative advisory roles show slower, more cautious adoption. |
| Augmentation potential | claude-haiku-4-5-20251001 | 4/5 | AI already assists economists significantly by rapidly synthesizing data, drafting sections of reports, stress-testing models, and surfacing relevant research—substantially raising analyst productivity while the economist retains judgment and client responsibility. |
| Augmentation potential | claude-sonnet-5 | 4/5 | AI substantially aids economists by accelerating literature review, data analysis, and report drafting, freeing time for the higher-value advisory judgment components. |
| Task automatability | claude-haiku-4-5-20251001 | 2/5 | While AI can generate economic analyses, market summaries, and preliminary policy briefs, the task fundamentally requires understanding client-specific contexts, advising on strategic tradeoffs, and delivering actionable recommendations tailored to organizational constraints—elements that demand human judgment and accountability beyond current LLM capabilities. |
| Task automatability | claude-sonnet-5 | 2/5 | Advisory consultation requires synthesizing client-specific context, judgment, and relationship trust that current AI cannot fully replicate end-to-end, though it can support research and drafting portions of this work.5o time savings across the full advisory task are not yet demonstrated at equal quality. |
| Adoption barriers | claude-haiku-4-5-20251001 | 4/5 | Clients (especially government agencies and regulated firms) require accountability, liability, and often prefer a named consultant who can testify or sign off on analysis; fiduciary duty and reputational risk create strong organizational and legal friction against full automation. |
| Adoption barriers | claude-sonnet-5 | 3/5 | No licensing requirement universally applies, but liability for bad advice, client preference for human judgment, and organizational trust create moderate friction against full substitution. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 2/5 | Economic consultation commands premium billing (often $200–500/hour) because it bundles research, analysis, and reputation risk; current AI inference costs are low but integration, fact-checking, and required human oversight add material expense, keeping all-in cost comparable to or exceeding junior economist rates. |
| Cost vs. human wage | claude-sonnet-5 | 2/5 | While AI can cheaply produce draft analyses, the human oversight, liability, and client trust-building needed for real consultation keep effective costs closer to human-comparable levels. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 2/5 | No deployed system reliably performs independent economic consultation at scale; LLMs can draft analyses but lack the real-time data integration, verification protocols, and accountability mechanisms that actual consulting firms use, and cannot handle the client relationship and accountability dimensions. |
| Technical feasibility today | claude-sonnet-5 | 2/5 | AI tools (chatbots, analytics platforms) can generate economic analysis and summaries, but no deployed product independently delivers professional consultation to clients reliably at scale. |
Formulate recommendations, policies, or plans to solve economic problems or to interpret markets.
29CI 25–32 · exposure 25 · augmentation 75 · importance 3.9/5 · click for rater detail
Formulate recommendations, policies, or plans to solve economic problems or to interpret markets.
29| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 2/5 | Adoption is slow. While some financial firms use ML for market analysis, most government agencies and policy institutions remain cautious, relying heavily on human economists for primary recommendation work. Pilots and analytical tools are widespread, but autonomous or near-autonomous policy formulation by AI is rare in practice. |
| Sector adoption velocity | claude-sonnet-5 | 3/5 | Economics and policy analysis sit within professional/financial services, a fast-adopting sector for AI drafting tools, though actual policy formulation remains human-led with AI as a research aid. |
| Augmentation potential | claude-haiku-4-5-20251001 | 4/5 | AI augmentation is strong here: tools can rapidly process large datasets, generate scenario analyses, flag outliers, and draft initial interpretations, substantially raising the productivity of an economist who retains responsibility for judgment and recommendation. The economist's role shifts toward synthesis and validation rather than data assembly and calculation. |
| Augmentation potential | claude-sonnet-5 | 4/5 | AI substantially helps economists by synthesizing data, generating scenario analyses, and drafting policy briefs, materially speeding the research and writing phases while humans retain final judgment. |
| Task automatability | claude-haiku-4-5-20251001 | 2/5 | While AI can generate economic analyses, draft policy summaries, and identify market trends from data, formulating high-stakes recommendations requires integrated judgment about trade-offs, stakeholder impacts, and contextual nuance that current systems handle inconsistently. An AI might automate 20–30% of the supporting analytical work but cannot reliably produce end-to-end recommendations meeting the 50% time-saving threshold. |
| Task automatability | claude-sonnet-5 | 2/5 | AI can draft policy options and summarize economic reasoning, but formulating credible, context-sensitive recommendations requires judgment, stakeholder knowledge, and accountability that current systems cannot reliably supply end-to-end. |
| Adoption barriers | claude-haiku-4-5-20251001 | 4/5 | Significant barriers exist: policy recommendations carry legal and reputational liability, requiring human accountability; regulatory bodies (central banks, legislatures) demand that a credentialed economist attest to major economic advice; and organizational practice strongly prefers human sign-off on recommendations that drive resource allocation or regulation. |
| Adoption barriers | claude-sonnet-5 | 3/5 | No formal licensing requirement, but institutional accountability, reputational risk, and the need for expert sign-off on policy advice create meaningful friction against pure automation. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 2/5 | AI inference for economic analysis is cheap, but the integration cost is high: economists must still validate outputs, rerun models, and reframe recommendations for decision-makers. The all-in cost of AI-assisted analysis often approaches or exceeds a junior economist's time on the same task when oversight and error correction are included. |
| Cost vs. human wage | claude-sonnet-5 | 2/5 | AI drafting is cheap, but the human oversight, validation, and domain expertise needed to vet recommendations keeps overall cost comparable to or only modestly below expert economist time. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 2/5 | No deployed AI product reliably formulates policy recommendations or market interpretations autonomously at production scale in government or institutional settings. Tools exist for data analysis and forecasting, but the synthesis into actionable policy requires human economists to validate assumptions, weigh competing models, and account for political/social context—tasks current systems do not do dependably. |
| Technical feasibility today | claude-sonnet-5 | 2/5 | LLMs are used experimentally to brainstorm policy language or summarize research, but no deployed product autonomously produces validated economic policy recommendations used in production decision-making. |
Conduct research on economic issues, and disseminate research findings through technical reports or scientific articles in journals.
29CI 25–32 · exposure 25 · augmentation 88 · importance 3.8/5 · click for rater detail
Conduct research on economic issues, and disseminate research findings through technical reports or scientific articles in journals.
29| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 2/5 | Adoption of AI in economic research remains slow and limited to augmentation (drafting, literature synthesis). Most research institutions and economists have not integrated AI agents into production workflows; adoption is still in pilot and experimental phases. |
| Sector adoption velocity | claude-sonnet-5 | 3/5 | Economics research, often housed in academia, government, and finance, has moderate AI tool adoption (e.g., for coding, literature synthesis) but full research pipeline automation is still in pilot/exploratory stages. |
| Augmentation potential | claude-haiku-4-5-20251001 | 4/5 | AI substantially assists economists by accelerating literature review, generating drafts, automating data wrangling, and providing writing feedback—measurably raising productivity on routine sub-tasks while the economist retains control over research design, interpretation, and novelty. |
| Augmentation potential | claude-sonnet-5 | 5/5 | AI substantially boosts economist productivity via literature synthesis, coding assistance, data cleaning, drafting text, and statistical analysis support, while the economist retains control over research design and interpretation. |
| Task automatability | claude-haiku-4-5-20251001 | 2/5 | Current AI cannot independently conduct original economic research requiring novel data collection, methodology design, and causal inference verification. However, AI can assist with literature review, data analysis scripting, and drafting sections, saving perhaps 20–30% of time but not meeting the 50% threshold for end-to-end performance at equal quality. |
| Task automatability | claude-sonnet-5 | 2/5 | AI can assist with literature review, data analysis, and drafting but the core work of formulating novel economic questions, designing rigorous studies, and interpreting findings with domain judgment still requires substantial human expertise; full end-to-end automation at equal quality is not achievable today. |
| Adoption barriers | claude-haiku-4-5-20251001 | 4/5 | Strong barriers exist: peer review and journal publication gates require human accountability; funding agencies and institutions mandate human researcher responsibility; professional credentials and liability for research integrity create legal and organizational friction against full automation. |
| Adoption barriers | claude-sonnet-5 | 3/5 | No licensing requirement to publish research, but peer review, institutional affiliation norms, and reputational/credibility requirements create meaningful friction against full AI substitution in authorship and dissemination. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 2/5 | AI inference is cheap, but the labor saved per task is modest because research quality and novelty cannot be automated; human economists must still do the core intellectual work. Integration and oversight costs are non-trivial relative to the time actually saved. |
| Cost vs. human wage | claude-sonnet-5 | 2/5 | AI can cut costs for literature review and coding/analysis tasks, but human economists' expertise, judgment, and accountability for publishable research remain costly and irreplaceable, keeping overall cost savings modest. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 2/5 | No deployed product reliably conducts independent economic research end-to-end. AI tools exist for writing assistance and data processing, but original research requires human judgment on hypothesis formation, model selection, and interpretation—areas where AI deployments remain narrow and error-prone. |
| Technical feasibility today | claude-sonnet-5 | 2/5 | Deployed tools (e.g., AI writing assistants, statistical software with AI features, literature search tools) support parts of the workflow, but no production system independently conducts original economic research and publishes peer-reviewed findings reliably. |
Develop economic guidelines and standards, and prepare points of view used in forecasting trends and formulating economic policy.
29CI 25–32 · exposure 25 · augmentation 75 · importance 3.8/5 · click for rater detail
Develop economic guidelines and standards, and prepare points of view used in forecasting trends and formulating economic policy.
29| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 2/5 | Policy institutions adopt AI for analytical support (data processing, scenario modeling) but retain human economists for guideline development and policy formulation; actual displacement in this specific task is minimal and slow given institutional conservatism and accountability requirements. |
| Sector adoption velocity | claude-sonnet-5 | 3/5 | Economics and policy institutions (finance, government, academia) are moderately fast adopters of AI for research assistance, though core forecasting and policy formulation remain human-led with pilots more common than production deployment. |
| Augmentation potential | claude-haiku-4-5-20251001 | 4/5 | AI substantially augments economists by accelerating literature review, generating multiple scenario analyses, automating data compilation, and drafting exploratory frameworks, allowing human economists to focus on judgment, stakeholder synthesis, and policy reasoning. |
| Augmentation potential | claude-sonnet-5 | 4/5 | AI substantially boosts economists' productivity by rapidly analyzing data, summarizing literature, and drafting scenarios, while the economist retains judgment over final guidelines and policy positions. |
| Task automatability | claude-haiku-4-5-20251001 | 2/5 | While AI can assist with data analysis, trend identification, and draft document preparation, developing genuine economic guidelines and formulating policy requires substantive judgment about complex systems, stakeholder tradeoffs, and normative values that current AI cannot perform end-to-end with 50% time savings at equal quality. |
| Task automatability | claude-sonnet-5 | 2/5 | This task requires original judgment, synthesis of political and social context, and accountability for policy positions, which current AI cannot independently perform end-to-end despite being able to draft supporting analysis.imestamp |
| Adoption barriers | claude-haiku-4-5-20251001 | 4/5 | Strong barriers exist: policy guidelines require human institutional authority and democratic accountability, central banks and treasuries legally require qualified economists to attest to forecasts, and liability for wrong guidance creates asymmetric error costs that deter full automation. |
| Adoption barriers | claude-sonnet-5 | 3/5 | No formal licensing requirement exists, but institutional credibility, reputational liability, and the need for expert accountability create meaningful friction against full automation. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 2/5 | Current AI tools are cheaper for individual analyses but require expensive human expert oversight and integration into institutional policy workflows; the total cost remains comparable to or higher than economist time when oversight and validation are included. |
| Cost vs. human wage | claude-sonnet-5 | 2/5 | While AI can cheaply produce drafts, the human oversight, expertise validation, and institutional accountability required keep all-in costs closer to comparable rather than order-of-magnitude cheaper. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 2/5 | No deployed AI system reliably produces standalone economic guidelines or policy positions; research tools and language models can draft components but lack the institutional credibility, legal authority, and judgment validation that real-world policy development demands. |
| Technical feasibility today | claude-sonnet-5 | 2/5 | AI tools can generate economic summaries and data analysis but no deployed product autonomously develops authoritative economic guidelines or policy positions used by institutions today. |
Teach theories, principles, and methods of economics.
26CI 25–28 · exposure 25 · augmentation 75 · importance 3.5/5 · click for rater detail
Teach theories, principles, and methods of economics.
26| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 2/5 | Universities and educational institutions are slow adopters; pilots with AI tutoring and content generation are increasing, but production displacement of teaching roles remains minimal. Budget constraints and governance structures limit rapid deployment. |
| Sector adoption velocity | claude-sonnet-5 | 3/5 | Higher education and academic economics are moderately adopting AI tools for course material generation and tutoring support, but full-scale replacement of instructors remains rare. |
| Augmentation potential | claude-haiku-4-5-20251001 | 4/5 | AI can meaningfully assist economists in drafting lecture materials, explaining difficult concepts through multiple framings, generating problem sets, and providing real-time student support, substantially raising educator productivity while the instructor remains the primary instructor and authority. |
| Augmentation potential | claude-sonnet-5 | 4/5 | AI can generate lecture materials, explain complex theories, create practice problems, and provide supplementary tutoring, meaningfully enhancing an economist's teaching productivity. |
| Task automatability | claude-haiku-4-5-20251001 | 2/5 | Teaching economics theory requires sustained engagement, responsiveness to student questions, and dynamic adaptation to learning needs that current AI struggles with end-to-end. While AI can draft lecture notes or explain concepts, it cannot reliably manage a full course's pedagogical flow, assessment, and real-time instruction with 50% time savings at equal learning outcomes. |
| Task automatability | claude-sonnet-5 | 2/5 | Teaching involves live interaction, adaptive pedagogy, student engagement, and assessment that current AI cannot fully replicate end-to-end, though it can assist with content generation and explanation.assistant |
| Adoption barriers | claude-haiku-4-5-20251001 | 4/5 | Accreditation bodies, institutions, and regulatory frameworks typically require a credentialed instructor to oversee and be accountable for course quality, grading, and student outcomes. Liability and duty-of-care expectations create hard barriers to full automation. |
| Adoption barriers | claude-sonnet-5 | 4/5 | Formal teaching roles in accredited institutions typically require credentialed instructors of record, and universities have strong organizational and accreditation requirements limiting full AI substitution. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 2/5 | AI infrastructure (model hosting, integration with LMS, oversight by educators) plus human oversight remains substantial relative to incremental teaching labor savings. Fully replacing an economist instructor's teaching load would require significant coordination costs. |
| Cost vs. human wage | claude-sonnet-5 | 2/5 | While AI content generation is cheap, replicating the full teaching function (interaction, grading nuance, institutional accreditation) still requires substantial human oversight, keeping cost savings limited. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 2/5 | No deployed product reliably teaches a full economics course independently; AI tutoring systems exist but are narrow, scaffold-heavy, and supplement rather than replace human instruction. Production systems use AI for content generation or tutoring sub-tasks, not end-to-end teaching. |
| Technical feasibility today | claude-sonnet-5 | 2/5 | AI tutoring products exist and can explain economic concepts, but no deployed system reliably substitutes for classroom teaching, mentorship, and interactive instruction at scale in academic institutions. |
Supervise research projects and students' study projects.
16CI 7–25 · exposure 13 · augmentation 50 · importance 3.8/5 · click for rater detail
Supervise research projects and students' study projects.
16| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 2/5 | Academic institutions have been slow to adopt AI systems for supervision roles, with most experiments limited to pilot feedback tools. The sector's emphasis on human mentorship, institutional inertia, and governance structures inhibit rapid deployment of autonomous or near-autonomous supervisory systems. |
| Sector adoption velocity | claude-sonnet-5 | 2/5 | While economists and academia use AI tools for research assistance, actual adoption of AI for supervisory/mentorship functions is minimal and not part of measured production deployment. |
| Augmentation potential | claude-haiku-4-5-20251001 | 3/5 | AI can usefully assist supervisors by summarizing student progress, flagging potential methodological issues in research proposals, and drafting feedback on written work. These augmentations can raise a supervisor's productivity in managing multiple projects, though the human supervisor remains in decision-making control. |
| Augmentation potential | claude-sonnet-5 | 3/5 | AI can help supervisors by drafting feedback, summarizing student work, or suggesting research directions, providing moderate assistance while the human retains supervisory responsibility. |
| Task automatability | claude-haiku-4-5-20251001 | 2/5 | While AI can assist with monitoring research timelines, flagging methodological issues, and providing feedback on drafts, the core supervisory responsibilities—making judgment calls on project direction, mentoring students through conceptual challenges, and ensuring research integrity—require human evaluation and accountability. AI cannot reliably replace the full supervisory relationship. |
| Task automatability | claude-sonnet-5 | 1/5 | Supervising research projects and mentoring students requires relational judgment, feedback, mentorship, and accountability that current AI cannot perform end-to-end.' |
| Adoption barriers | claude-haiku-4-5-20251001 | 4/5 | Academic institutions typically require qualified faculty to supervise research projects and students, often stipulated by funding bodies, accreditation standards, and institutional policy. Liability, mentorship responsibility, and research integrity sign-off are legally or contractually bound to human supervisors. |
| Adoption barriers | claude-sonnet-5 | 4/5 | Institutional accreditation, academic advising responsibilities, and mentorship expectations tied to human faculty roles create strong organizational and credentialing barriers to substitution. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 2/5 | The cost of integrating AI monitoring systems, training supervisors to use them, and the overhead of human oversight of AI feedback still exceeds the value of partial automation. Human supervisors remain the cheaper, trusted option for this complex interpersonal and accountability role. |
| Cost vs. human wage | claude-sonnet-5 | 1/5 | There is no viable AI substitute performing this supervisory role, so the cost comparison favors the human by default since AI cannot deliver the output. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 2/5 | Current AI systems can generate feedback on research documents and highlight potential issues, but no deployed product reliably supervises student projects end-to-end or makes substantive decisions about research direction. Existing tools serve as assistants, not replacements for human supervisors. |
| Technical feasibility today | claude-sonnet-5 | 1/5 | No deployed product autonomously supervises human researchers or students; this remains a human-managed, research-stage-only scenario for AI in a supervisory capacity. |
Provide litigation support, such as writing reports for expert testimony or testifying as an expert witness.
15CI 15–15 · exposure 16 · augmentation 63 · importance 3.2/5 · click for rater detail
Provide litigation support, such as writing reports for expert testimony or testifying as an expert witness.
15| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 2/5 | Adoption remains slow because litigation is highly regulated, risk-averse, and dependent on professional credentials and courtroom presence. Firms use AI drafting assistants, but substantive displacement of economist testimony has not materialized in production legal workflows. |
| Sector adoption velocity | claude-sonnet-5 | 2/5 | Legal and litigation support sectors are cautious with AI due to evidentiary and ethical concerns, with adoption limited mostly to research assistance and document review rather than testimony generation. |
| Augmentation potential | claude-haiku-4-5-20251001 | 3/5 | AI assists economists by accelerating literature review, data synthesis, and initial report drafting, raising the speed of evidence assembly. However, the human economist retains full responsibility for methodology, conclusions, and testimony, so augmentation is meaningful but bounded by the need for human judgment and credibility. |
| Augmentation potential | claude-sonnet-5 | 4/5 | AI tools can meaningfully speed up literature review, data analysis, and drafting portions of expert reports, letting economists focus on judgment-intensive analysis and testimony preparation. |
| Task automatability | claude-haiku-4-5-20251001 | 2/5 | While AI can draft report components and summarize economic evidence, expert testimony requires credible professional judgment under cross-examination and legal accountability that current AI systems cannot independently deliver. An economist must synthesize complex evidence, defend methodology, and adapt to legal context in real-time—tasks where AI support is substantial but autonomous end-to-end automation falls far short of the 50% time-saving threshold. |
| Task automatability | claude-sonnet-5 | 2/5 | AI can help draft portions of reports and analyze data, but synthesizing case-specific economic analysis, exercising professional judgment, and withstanding cross-examination requires human expertise; the full task cannot be automated end-to-end today. |
| Adoption barriers | claude-haiku-4-5-20251001 | 5/5 | Legal and professional requirements mandate that an accredited economist author and testify to expert reports; courts require a qualified human witness for cross-examination and accountability. Licensing, liability, and the legal definition of expert testimony are hard barriers to automation. |
| Adoption barriers | claude-sonnet-5 | 5/5 | Expert witness testimony requires a qualified, credentialed human to be sworn in, cross-examined, and legally accountable for opinions offered under oath—an unautomatable legal requirement. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 2/5 | Expert witness fees ($300–$1,000+ per hour) remain high due to liability, reputation, and credential requirements; AI can reduce report-drafting time but cannot displace the expert's fee-billable presence in deposition and testimony. Cost savings are modest relative to total engagement cost. |
| Cost vs. human wage | claude-sonnet-5 | 2/5 | AI can cheaply generate draft text or run statistical models, but the litigation support task requires a credentialed expert's sworn testimony and legal accountability, so overall cost is dominated by human expert fees regardless of AI assistance. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 1/5 | No deployed product reliably performs expert witness testimony or produces courtroom-admissible expert reports autonomously; these require a licensed economist to author, vouch for, and defend findings. AI tools exist to assist drafting, but production systems do not substitute for the human expert's legal and professional standing. |
| Technical feasibility today | claude-sonnet-5 | 1/5 | No deployed product testifies as an expert witness, and using AI-drafted reports without extensive human validation risks credibility and admissibility challenges in court; this remains research/assistive-stage only. |
Testify at regulatory or legislative hearings concerning the estimated effects of changes in legislation or public policy, and present recommendations based on cost-benefit analyses.
1CI 0–3 · exposure 0 · augmentation 63 · importance 3.3/5 · click for rater detail
Testify at regulatory or legislative hearings concerning the estimated effects of changes in legislation or public policy, and present recommendations based on cost-benefit analyses.
1| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 1/5 | Adoption of AI replacement is not occurring because legal and regulatory frameworks mandate human testimony and expert accountability. Sectors (government, legislature) are slow to digitize and fundamentally require human expert judgment and presence. |
| Sector adoption velocity | claude-sonnet-5 | 2/5 | Government and policy analysis sectors are slower adopters of AI-driven decision presentation, and hearing testimony itself has seen no displacement. |
| Augmentation potential | claude-haiku-4-5-20251001 | 3/5 | AI can assist by preparing cost-benefit analyses, synthesizing data, and drafting talking points or testimony outlines that the economist then refines and delivers. This is moderately helpful but the core task—persuasive, accountable testimony—remains fundamentally human-dependent. |
| Augmentation potential | claude-sonnet-5 | 4/5 | AI can strongly assist economists in drafting cost-benefit analyses, modeling policy impacts, and preparing testimony content, though the human must still deliver and defend it. |
| Task automatability | claude-haiku-4-5-20251001 | 1/5 | This task requires real-time testimony, cross-examination, political judgment, and credibility-bearing presentation of analyses to legislators or regulators. AI cannot provide sworn testimony, respond to hostile questioning, or bear accountability for policy recommendations in a legal/institutional context. The human expert must be present and accountable. |
| Task automatability | claude-sonnet-5 | 1/5 | Testifying in person before legislative or regulatory bodies requires live human presence, credibility, and real-time judgment that AI cannot perform end-to-end today. |
| Adoption barriers | claude-haiku-4-5-20251001 | 5/5 | Hard legal and institutional barriers: testimony must be given by an identified, qualified human expert who can be cross-examined, sworn in, and held accountable for statements. Regulatory and legislative processes require human accountability and expert standing that AI cannot assume. |
| Adoption barriers | claude-sonnet-5 | 5/5 | Legislative and regulatory hearings require a credentialed, accountable human witness who can be questioned, sworn, and held responsible—an essentially hard institutional and legal barrier. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 1/5 | The value of expert testimony lies in the economist's credibility, standing, and legal accountability—factors AI cannot provide. The cost of AI-assisted preparation would be minimal, but the core task (testifying) requires the human expert, making automation economically moot. |
| Cost vs. human wage | claude-sonnet-5 | 1/5 | There is no AI substitute for the testimony act itself, so cost comparison favors the human by default even though AI can help prepare materials cheaply. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 1/5 | No deployed AI system can legally or institutionally substitute for expert testimony; regulatory bodies and legislative committees require identified human experts who can be held accountable. AI might assist in drafting analyses, but cannot testify or represent expert conclusions to decision-makers. |
| Technical feasibility today | claude-sonnet-5 | 1/5 | No deployed product testifies at hearings; this remains firmly a human institutional role with no production AI substitute. |
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.