Actuaries
15-2011.00Median wage $130,000/yr26,670 employed (US)Rank #370 of 923 scored · top 40% by substitution
Analyze statistical data, such as mortality, accident, sickness, disability, and retirement rates and construct probability tables to forecast risk and liability for payment of future benefits. May ascertain insurance rates required and cash reserves necessary to ensure payment of future benefits.
Sub-scores
0–100 · band = confidence interval from rater disagreement
Substitution31
Exposure31
Augmentation70
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
15 rated tasks, binned by substitution score.
Position among all scored occupations
Distribution of 923 occupation scores; the marker is this occupation.
Tasks with substitution ≥ 70
0%
Run 1.0.0-draft.1 · computed 2026-08-05 · rater panel: claude-sonnet-5, claude-haiku-4-5-20251001 · intervals span rater disagreement.
Why this score
The five weighted dimensions of the composite, averaged across this occupation's tasks (importance-weighted, panel mean). Exact weights and formulas: /api/v1/methodology.
Task automatabilityw 35%33
panel mean rating 2.3/5 → substitution pressure 33/100
Technical feasibility todayw 20%28
panel mean rating 2.1/5 → substitution pressure 28/100
Cost vs. human wagew 15%35
panel mean rating 2.4/5 → substitution pressure 35/100
Adoption barriersw 20%inverted — strong barriers lower the score23
panel mean rating 4.1/5 (barrier strength) → substitution pressure 23/100
Sector adoption velocityw 10%38
panel mean rating 2.5/5 → substitution pressure 38/100
Task breakdown (15 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.
Analyze statistical information to estimate mortality, accident, sickness, disability, and retirement rates.
56CI 45–67 · exposure 62 · augmentation 75 · importance 4.4/5 · click for rater detail
Analyze statistical information to estimate mortality, accident, sickness, disability, and retirement rates.
56| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 3/5 | Insurance and pension sectors have adopted predictive modeling and machine learning for claims forecasting, but adoption remains cautious and mixed. Pilots are common; full autonomous replacement is rare due to regulatory caution and the importance of professional judgment in assumption-setting. |
| Sector adoption velocity | claude-sonnet-5 | 3/5 | Insurance and finance sectors are moderately fast adopters of AI/statistical tools, with many pilots and some production use in underwriting and risk modeling, though full task automation remains uncommon. |
| Augmentation potential | claude-haiku-4-5-20251001 | 4/5 | AI substantially enhances actuarial productivity by automating data wrangling, fitting multiple candidate models, sensitivity analysis, and backtesting—letting human actuaries focus on assumption logic, model selection, and regulatory communication. This human-in-the-loop pattern is already mainstream. |
| Augmentation potential | claude-sonnet-5 | 4/5 | AI and advanced statistical/ML tools substantially speed up data analysis, model fitting, and scenario testing, meaningfully boosting actuary productivity while judgment and sign-off remain human-driven. |
| Task automatability | claude-haiku-4-5-20251001 | 4/5 | Current AI systems can collect, preprocess, and analyze large statistical datasets to estimate mortality and other rates with high quality and speed. While the task requires modeling choices and assumption validation that benefit from human oversight, the core statistical estimation—fitting GLMs, neural networks, or ensemble models to historical claims data—is automatable at scale and delivers time savings well above 50%. |
| Task automatability | claude-sonnet-5 | 3/5 | AI can run statistical models and generate rate estimates quickly, but the task requires domain judgment, regulatory context, and validation that current systems cannot fully replace end-to-end without significant human oversight and setup. |
| Adoption barriers | claude-haiku-4-5-20251001 | 3/5 | Actuarial work is subject to professional licensing (ASA, FSA, EA certifications) and regulatory oversight (state insurance boards, SOX, Solvency II). While AI can perform the statistical analysis, formal sign-off often requires a credentialed human, creating moderate friction but not an absolute legal prohibition on AI execution. |
| Adoption barriers | claude-sonnet-5 | 4/5 | Actuarial work is subject to professional certification requirements (e.g., signing off on reserves/reports) and regulatory scrutiny, creating strong liability and licensing barriers to full automation. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 4/5 | Inference and model serving for statistical estimation is cheap compared to the loaded cost of an actuary (typically $150k–250k+ annually). Once trained, model retraining and inference on new data costs a small fraction of human labor, especially for routine batch updates. |
| Cost vs. human wage | claude-sonnet-5 | 3/5 | AI-assisted statistical modeling can reduce analyst hours, but licensed actuaries, data infrastructure, and model validation still require significant paid human time, keeping costs roughly comparable rather than dramatically cheaper. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 4/5 | Deployed actuarial software and AI-powered analytics platforms (e.g., insurance firms using machine learning pipelines, predictive modeling tools) demonstrably perform mortality and claims-rate estimation in production. Some gaps remain around explainability and regulatory sign-off, but the underlying statistical inference is reliable and widely implemented. |
| Technical feasibility today | claude-sonnet-5 | 3/5 | Deployed actuarial software with statistical/ML components exists and is used for parts of this analysis, but fully autonomous, reliable production systems performing the entire estimation task independently are not standard. |
Construct probability tables for events such as fires, natural disasters, and unemployment, based on analysis of statistical data and other pertinent information.
55CI 36–74 · exposure 58 · augmentation 88 · importance 4.2/5 · click for rater detail
Construct probability tables for events such as fires, natural disasters, and unemployment, based on analysis of statistical data and other pertinent information.
55| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 4/5 | Actuarial and insurance sectors are high-digitization, data-driven industries with strong incentives to automate. Major insurers and pension funds already deploy AI and ML for risk modeling; table construction automation follows established patterns in financial services adoption. |
| Sector adoption velocity | claude-sonnet-5 | 3/5 | Insurance and financial services are moderately fast adopters of AI/ML for risk modeling, with many pilots and some production use, but full automation of core actuarial deliverables remains cautious due to regulatory scrutiny. |
| Augmentation potential | claude-haiku-4-5-20251001 | 5/5 | AI dramatically augments actuaries by automating data prep, suggesting probability distributions, and generating candidate tables, leaving actuaries to validate assumptions, interpret results, and ensure regulatory compliance. This transforms productivity while keeping humans central to judgment and sign-off. |
| Augmentation potential | claude-sonnet-5 | 4/5 | AI and advanced statistical/ML tools significantly speed up data analysis, pattern detection, and scenario modeling, substantially boosting actuarial productivity even though humans remain responsible for final tables and judgment calls. |
| Task automatability | claude-haiku-4-5-20251001 | 4/5 | AI systems can analyze statistical data, identify patterns, and construct probability tables with high accuracy and minimal human oversight. Modern ML and statistical tools can perform the end-to-end pipeline—data ingestion, outlier detection, distribution fitting, and table generation—substantially faster than manual construction, meeting the ≥50% time-saving threshold. |
| Task automatability | claude-sonnet-5 | 3/5 | AI can perform much of the statistical modeling and data analysis underlying probability table construction, but requires actuarial judgment, domain-specific adjustments, and validation that current systems cannot fully replace end-to-end without significant human oversight. |
| Adoption barriers | claude-haiku-4-5-20251001 | 3/5 | Regulatory requirements (insurance regulators often demand sign-off by a credentialed actuary) and liability concerns create meaningful friction. The final probability tables may need human actuary approval before deployment, reducing pure automation but not blocking AI assistance in the construction phase. |
| Adoption barriers | claude-sonnet-5 | 4/5 | Actuarial work is often subject to professional certification (e.g., Fellow of the Society of Actuaries), regulatory review, and legal liability for pricing/reserving decisions, creating strong barriers to full automation. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 5/5 | AI-driven table construction costs orders of magnitude less than hiring an actuary to manually compile and analyze data. Once trained on historical patterns, inference is near-zero cost, and the human oversight required is minimal compared to the full manual workflow. |
| Cost vs. human wage | claude-sonnet-5 | 2/5 | AI can cheaply generate statistical estimates, but the cost of required actuarial review, validation, and compliance checks keeps the all-in cost closer to comparable rather than dramatically cheaper. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 4/5 | Deployed products (statistical software, data platforms, and ML frameworks) reliably construct probability tables from structured data in production environments. Specialized actuarial software and general-purpose data science tools (Python, R, Spark) have mature implementations, though domain-specific validation and interpretation still often require human review. |
| Technical feasibility today | claude-sonnet-5 | 2/5 | While statistical/ML tools are widely used in actuarial work, deployed AI products that autonomously construct validated probability tables for regulatory or pricing use are not yet standard production practice; humans remain central to model construction and sign-off. |
Ascertain premium rates required and cash reserves and liabilities necessary to ensure payment of future benefits.
50CI 23–78 · exposure 58 · augmentation 75 · importance 4.5/5 · click for rater detail
Ascertain premium rates required and cash reserves and liabilities necessary to ensure payment of future benefits.
50| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 4/5 | Insurance and pension sectors are highly digitized, regulated, and competitive; they have rapidly adopted actuarial software and increasingly AI-augmented tools over the past decade. Large organizations deploy these systems extensively, though smaller firms and certain niche segments lag behind. |
| Sector adoption velocity | claude-sonnet-5 | 3/5 | Insurance and financial services are moderately fast adopters of AI for underwriting and analytics, but core reserving/pricing sign-off remains conservative and heavily regulated, slowing full production deployment. |
| Augmentation potential | claude-haiku-4-5-20251001 | 4/5 | AI and software tools significantly amplify actuarial productivity by automating routine calculations, scenario modeling, and sensitivity analysis, freeing actuaries to focus on assumption-setting, model validation, and judgment-heavy decisions. Humans remain in the loop for sign-off and complex interpretive work. |
| Augmentation potential | claude-sonnet-5 | 4/5 | AI substantially assists actuaries via predictive modeling, automated data aggregation, and scenario analysis, meaningfully speeding up rate and reserve calculations while the actuary retains final judgment. |
| Task automatability | claude-haiku-4-5-20251001 | 5/5 | This task involves deterministic mathematical calculations on structured data (mortality tables, interest rates, policy terms) to compute premium rates and reserve amounts. Modern AI systems and specialized actuarial software can perform these computations end-to-end with >50% time savings at equal quality by automating data extraction, model selection, and reserve calculation. |
| Task automatability | claude-sonnet-5 | 2/5 | Core actuarial judgment involves regulatory compliance, model selection, and professional certification that current AI cannot fully replace end-to-end, though it can accelerate calculations and data preparation within the workflow. |
| Adoption barriers | claude-haiku-4-5-20251001 | 4/5 | Significant regulatory barriers exist: actuaries must sign off on valuations for solvency and statutory reporting in most jurisdictions, and professional liability attaches to the actuary's signature, not the software. Organizations typically require human actuarial review and certification, creating a hard legal requirement that prevents full substitution. |
| Adoption barriers | claude-sonnet-5 | 5/5 | Actuarial certification (e.g., Appointed Actuary opinions) is legally mandated in insurance regulation, requiring a licensed professional to sign off on reserve adequacy and rate filings. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 5/5 | AI inference and actuarial software operate at negligible cost per calculation once implemented, while actuaries command six-figure salaries. The all-in cost per reserve/premium computation is typically one to two orders of magnitude lower with automation than with human actuaries performing the same calculations. |
| Cost vs. human wage | claude-sonnet-5 | 2/5 | While computational modeling can be cheaper than manual calculation, the requirement for credentialed actuarial review and validation keeps overall cost comparable to or only modestly below human-driven processes. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 4/5 | Mature actuarial software and increasingly AI-enhanced tools perform reserve and premium calculations reliably in production environments at major insurers and pension funds. Some edge cases (novel product structures, regulatory interpretation) still require human oversight, but core computational tasks are deployed at scale. |
| Technical feasibility today | claude-sonnet-5 | 2/5 | AI/ML tools are used for pricing analytics and reserving support in production, but no deployed product independently ascertains final premium rates or statutory reserves without actuarial sign-off. |
Determine policy contract provisions for each type of insurance.
42CI 28–57 · exposure 45 · augmentation 75 · importance 3.0/5 · click for rater detail
Determine policy contract provisions for each type of insurance.
42| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 3/5 | Insurance and financial services are moderately early in LLM and agent adoption; some insurers are piloting contract automation, but widespread production deployment of AI-generated policy provisions remains limited by compliance and risk-aversion norms. |
| Sector adoption velocity | claude-sonnet-5 | 3/5 | Insurance is a digitizing but conservative, heavily regulated sector; AI pilots for document drafting and analysis are growing but full contract determination remains rare in production. |
| Augmentation potential | claude-haiku-4-5-20251001 | 4/5 | AI drafting and comparison tools can substantially accelerate an actuary's work on contract review and provision assembly, helping them explore alternatives and spot inconsistencies faster while maintaining their judgment on complex or novel provisions. |
| Augmentation potential | claude-sonnet-5 | 4/5 | AI can meaningfully speed up drafting, benchmarking against existing provisions, and identifying regulatory or risk issues, substantially aiding actuaries while they retain final judgment. |
| Task automatability | claude-haiku-4-5-20251001 | 4/5 | AI systems can now draft, review, and suggest policy contract provisions by analyzing existing templates, regulatory requirements, and risk data with high consistency, reducing human review time by >50%. However, final legal sign-off and negotiation of complex, non-standard provisions remain human-dependent. |
| Task automatability | claude-sonnet-5 | 2/5 | Determining contract provisions requires integrating regulatory constraints, actuarial risk assessment, and business strategy judgment that current AI cannot fully replicate end-to-end, though it can draft and suggest language. |
| Adoption barriers | claude-haiku-4-5-20251001 | 4/5 | Insurance is heavily regulated; policy contract provisions must often be reviewed and approved by licensed actuaries or legal counsel, and errors carry significant liability exposure, creating material friction against full automation. |
| Adoption barriers | claude-sonnet-5 | 4/5 | Insurance contract provisions are subject to regulatory filing requirements and require sign-off by licensed actuaries/legal counsel, creating strong institutional and liability barriers to full automation. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 4/5 | AI inference and integration costs are a small fraction of the loaded cost of an actuary's time spent drafting or reviewing standard policy provisions, though oversight and validation add some overhead. |
| Cost vs. human wage | claude-sonnet-5 | 2/5 | Given the need for extensive human actuarial and legal oversight to validate AI-suggested provisions, cost savings are modest rather than order-of-magnitude. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 3/5 | LLM-based contract generation and analysis tools exist in production (e.g., legal tech platforms), but they are typically narrower in scope than full policy contract determination and still require material human review for accuracy and compliance in regulated insurance contexts. |
| Technical feasibility today | claude-sonnet-5 | 2/5 | AI tools can retrieve and compare policy language and flag inconsistencies, but no deployed product independently determines final contract provisions in production without heavy actuarial/legal review. |
Determine, or help determine, company policy, and explain complex technical matters to company executives, government officials, shareholders, policyholders, or the public.
40CI 18–62 · exposure 41 · augmentation 75 · importance 4.3/5 · click for rater detail
Determine, or help determine, company policy, and explain complex technical matters to company executives, government officials, shareholders, policyholders, or the public.
40| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 3/5 | Large insurance and financial firms are actively piloting AI-assisted policy drafting and executive communication tools, but full production deployment for policy determination remains measured and cautious due to reputational and regulatory risk. |
| Sector adoption velocity | claude-sonnet-5 | 3/5 | Insurance and financial services are moderately fast adopters of AI for analytics and drafting support, but governance and executive-facing decision authority remain human-controlled with slow structural change. |
| Augmentation potential | claude-haiku-4-5-20251001 | 4/5 | AI excels at rapidly generating multiple policy framings, translating complex technical actuarial concepts into executive-friendly summaries, and identifying logical gaps in proposed policies; these augment human decision-makers substantially while keeping them in control. |
| Augmentation potential | claude-sonnet-5 | 4/5 | AI can strongly assist by summarizing technical models, drafting stakeholder communications, and simplifying complex actuarial concepts for non-technical audiences, boosting actuary productivity substantially. |
| Task automatability | claude-haiku-4-5-20251001 | 4/5 | AI systems can draft policy documents, synthesize technical analyses, and generate clear executive summaries with significant time savings; however, the final determination of policy and the judgment calls about competitive strategy and legal/regulatory implications still require human oversight, preventing a full 5-rating. |
| Task automatability | claude-sonnet-5 | 2/5 | This task blends strategic judgment, organizational authority, and stakeholder communication that current AI cannot autonomously perform end-to-end, though drafting explanations is assistable.' |
| Adoption barriers | claude-haiku-4-5-20251001 | 3/5 | Insurance and pension policy decisions carry regulatory scrutiny and fiduciary duties that create some friction; however, there is no strict legal requirement that a human must personally author policy statements, only that they be reviewed and approved by qualified professionals. |
| Adoption barriers | claude-sonnet-5 | 5/5 | Actuarial sign-off, fiduciary responsibility, and regulatory requirements mean licensed professionals must make and be accountable for these determinations and communications. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 4/5 | AI inference and document generation costs are now an order of magnitude lower than the loaded wage of a senior actuarial professional, even accounting for oversight and integration overhead. |
| Cost vs. human wage | claude-sonnet-5 | 2/5 | AI can cheaply generate explanatory drafts, but the actual policy-setting and high-stakes communication still requires expensive human expert time and accountability, keeping overall cost comparable to human-led work. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 3/5 | LLM-based tools and agent systems can produce competent first drafts of policy briefs and technical explanations at scale, but organizations rarely deploy end-to-end AI decision-making on material policy matters; human review and sign-off remain standard practice. |
| Technical feasibility today | claude-sonnet-5 | 1/5 | No deployed product independently determines company policy or represents actuarial conclusions to executives/regulators; this remains a human executive/professional function. |
Explain changes in contract provisions to customers.
39CI 25–54 · exposure 38 · augmentation 75 · importance 2.4/5 · click for rater detail
Explain changes in contract provisions to customers.
39| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 2/5 | Insurance and actuarial firms are cautious adopters of fully automated customer-facing explanations due to regulatory scrutiny, professional certification requirements, and risk aversion around contract misstatement; pilots exist but production deployment remains limited. |
| Sector adoption velocity | claude-sonnet-5 | 3/5 | Insurance and financial services are adopting AI customer service tools at a moderate pace, with pilots widespread but full-scale reliance on AI for regulated explanations still limited. |
| Augmentation potential | claude-haiku-4-5-20251001 | 4/5 | AI assistants can effectively draft clear language summaries, highlight key changes, and propose answers to common questions, significantly raising actuary productivity while the licensed professional remains responsible for accuracy, customization, and regulatory sign-off. |
| Augmentation potential | claude-sonnet-5 | 4/5 | AI can significantly help actuaries and customer-facing staff draft clear, consistent explanations of contract changes, improving speed and consistency while humans retain accountability. |
| Task automatability | claude-haiku-4-5-20251001 | 2/5 | While AI can generate draft explanations of contract terms, the task requires tailoring complex language to individual customer contexts, addressing their concerns, and adapting dynamically to questions—elements that typically require human judgment and real-time interaction to meet the ≥50% time-saving threshold reliably. |
| Task automatability | claude-sonnet-5 | 3/5 | AI chatbots and LLM-based tools can draft or deliver explanations of contract changes in plain language, but nuanced customer-specific interpretation and trust-building still often require human involvement, so only partial time savings are realized end-to-end.dollar |
| Adoption barriers | claude-haiku-4-5-20251001 | 4/5 | Strong regulatory and professional barriers exist: actuaries must often formally certify explanations, customers may require licensed communication, and errors in contract explanation carry material legal and financial liability that organizations are reluctant to delegate fully to automation. |
| Adoption barriers | claude-sonnet-5 | 3/5 | Insurance communications are regulated for accuracy and fairness, and some jurisdictions require licensed personnel for certain explanations, creating moderate compliance and liability friction. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 2/5 | Current AI inference plus integration and oversight (risk of misstatement in financial/legal contexts) remains comparable to or more expensive than a trained actuary or customer service representative when accounting for error correction and liability exposure. |
| Cost vs. human wage | claude-sonnet-5 | 4/5 | Automated explanation systems (chat/IVR/LLM-driven support) are far cheaper per interaction than having an actuary or licensed agent personally explain changes, though oversight costs remain. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 2/5 | No deployed product reliably performs end-to-end customer-facing contract explanation in production at scale; AI can draft explanations but struggles with domain-specific nuance, liability concerns, and the interactive nature of customer communication in insurance contexts. |
| Technical feasibility today | claude-sonnet-5 | 3/5 | Insurance chatbots and customer service AI already explain policy terms in production, but accuracy on complex actuarial contract provisions is inconsistent and often escalated to humans. |
Collaborate with programmers, underwriters, accounts, claims experts, and senior management to help companies develop plans for new lines of business or improvements to existing business.
29CI 25–32 · exposure 25 · augmentation 63 · importance 4.5/5 · click for rater detail
Collaborate with programmers, underwriters, accounts, claims experts, and senior management to help companies develop plans for new lines of business or improvements to existing business.
29| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 2/5 | While financial and insurance sectors digitize quickly, strategic planning and cross-functional collaboration remain human-led activities. AI adoption in this space is limited to supporting tools (analytics, document prep) rather than agent-driven decision-making, reflecting slow displacement. |
| Sector adoption velocity | claude-sonnet-5 | 3/5 | Insurance and financial services are moderately fast adopters of AI for analytics, but strategic cross-departmental planning remains largely human-led with AI in a pilot/support role. |
| Augmentation potential | claude-haiku-4-5-20251001 | 3/5 | AI can meaningfully assist actuaries by automating data aggregation, generating scenario models, and drafting analysis—raising their output speed—while the actuary retains judgment and stakeholder coordination responsibilities. |
| Augmentation potential | claude-sonnet-5 | 4/5 | AI can significantly aid actuaries by generating data analyses, market scenarios, and draft proposals that inform collaborative discussions, enhancing productivity while humans retain decision-making and coordination roles. |
| Task automatability | claude-haiku-4-5-20251001 | 2/5 | While AI can assist in data analysis and modeling components, the task fundamentally requires cross-functional human collaboration, judgment on business strategy, and negotiation with multiple stakeholders. AI cannot currently conduct end-to-end strategic planning meetings or synthesize diverse expert inputs into cohesive business decisions. |
| Task automatability | claude-sonnet-5 | 2/5 | This is a cross-functional collaboration and strategic planning task involving synthesis of diverse expert input and organizational judgment, which current AI cannot autonomously replicate end-to-end. |
| Adoption barriers | claude-haiku-4-5-20251001 | 4/5 | Regulatory requirements (actuarial credentials, sign-off authority), fiduciary responsibility, and the legal necessity of qualified human actuaries in business planning decisions create strong barriers to full automation. Senior management and underwriting expertise cannot be delegated to AI agents alone. |
| Adoption barriers | claude-sonnet-5 | 3/5 | While no formal licensing is required for this specific collaborative task, actuarial sign-off and organizational trust in senior judgment create real friction against full automation. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 2/5 | An actuary's loaded compensation is substantial ($120k–$180k+), and AI tools do not yet provide sufficient strategic value to justify replacement costs when factoring in integration, oversight, and the residual need for human judgment and stakeholder management. |
| Cost vs. human wage | claude-sonnet-5 | 2/5 | Because the task requires ongoing human coordination and judgment, AI can only reduce some analytical workload, not replace the collaborative process, so cost savings are modest relative to the human labor still required. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 2/5 | No deployed product reliably performs collaborative strategy development across multiple departments. AI can support individual components (data analysis, document drafting) but cannot manage the interpersonal coordination, consensus-building, and strategic judgment this task demands in production settings. |
| Technical feasibility today | claude-sonnet-5 | 2/5 | AI tools can support parts of business planning (data analysis, drafting proposals) but no deployed product manages multi-stakeholder collaboration and strategic decision-making for new business lines reliably today. |
Design, review, and help administer insurance, annuity and pension plans, determining financial soundness and calculating premiums.
26CI 25–28 · exposure 25 · augmentation 75 · importance 4.4/5 · click for rater detail
Design, review, and help administer insurance, annuity and pension plans, determining financial soundness and calculating premiums.
26| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 2/5 | Financial services and insurance are moderate adopters of AI overall, but actuarial practices remain conservative and slow to adopt automation due to liability, regulation, and the specialized expertise required. Adoption is mostly in support tools, not replacement of core actuarial roles. |
| Sector adoption velocity | claude-sonnet-5 | 3/5 | Insurance and finance are moderately fast adopters of AI for modeling and analytics, but actuarial plan design remains conservative due to regulatory and liability concerns. |
| Augmentation potential | claude-haiku-4-5-20251001 | 4/5 | AI systems demonstrably assist actuaries through automated data processing, scenario modeling, and premium calculation, significantly raising productivity on analytical and computational parts of the work while the actuary retains design and judgment responsibilities. |
| Augmentation potential | claude-sonnet-5 | 4/5 | AI substantially aids actuaries in data analysis, scenario modeling, and premium calculation, improving efficiency while the actuary retains responsibility for design and certification. |
| Task automatability | claude-haiku-4-5-20251001 | 2/5 | While AI can assist with calculations and data analysis components, the task requires complex judgment about financial soundness, regulatory compliance, and plan design that demands human expertise. Current systems cannot reliably replace the end-to-end design and administration without significant human oversight. |
| Task automatability | claude-sonnet-5 | 2/5 | Plan design and financial soundness determination require complex regulatory, actuarial judgment and stakeholder negotiation that current AI cannot fully replicate end-to-end, though calculation-heavy sub-steps can be automated. |
| Adoption barriers | claude-haiku-4-5-20251001 | 4/5 | Actuarial work is heavily regulated; actuaries must be credentialed (ASA, FSA, EA designations) and legally accountable for plan soundness and compliance. Regulatory frameworks explicitly require qualified human actuaries to sign off on pension and insurance plans, creating hard barriers to substitution. |
| Adoption barriers | claude-sonnet-5 | 4/5 | Actuarial certification (e.g., Fellow/Associate designations) and regulatory requirements for signed actuarial opinions on plan soundness create strong legal barriers to full automation. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 2/5 | AI tools for actuarial work (modeling, data processing) remain specialized and expensive relative to the work they automate, while actuaries command high salaries but perform work that justifies that cost through judgment and liability. Integration overhead is substantial. |
| Cost vs. human wage | claude-sonnet-5 | 2/5 | AI can cheaply run calculations, but the human oversight, regulatory sign-off, and judgment needed for plan design and soundness review keep overall costs comparable to or only modestly below human-only costs. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 2/5 | No deployed product reliably performs the full scope of design, review, and administration of insurance plans independently. AI tools exist for premium calculation and risk modeling but fall short of handling the holistic plan design and regulatory sign-off that define actuarial work. |
| Technical feasibility today | claude-sonnet-5 | 2/5 | Some actuarial software and AI tools assist with modeling and premium calculations, but no deployed product independently designs and administers full plans reliably at scale. |
Provide expertise to help financial institutions manage risks and maximize returns associated with investment products or credit offerings.
26CI 25–28 · exposure 25 · augmentation 75 · importance 3.1/5 · click for rater detail
Provide expertise to help financial institutions manage risks and maximize returns associated with investment products or credit offerings.
26| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 2/5 | Financial institutions are experimenting with AI-driven analytics and risk tools, but adoption remains primarily in augmentation and component-level tasks (e.g., data preprocessing, scenario generation) rather than displacing actuarial advisory roles. Pilot-stage rather than deep production deployment dominates. |
| Sector adoption velocity | claude-sonnet-5 | 3/5 | Financial services is a fast-adopting sector for AI tools generally, but actuarial risk advisory specifically still shows mostly pilot-stage integration rather than full production reliance. |
| Augmentation potential | claude-haiku-4-5-20251001 | 4/5 | AI significantly enhances actuarial productivity by automating data aggregation, running stress tests, generating model scenarios, and flagging anomalies—freeing actuaries to focus on judgment, strategy, and client communication while remaining firmly in the decision-making loop. |
| Augmentation potential | claude-sonnet-5 | 4/5 | AI significantly enhances actuaries' work through faster data analysis, scenario modeling, and report drafting, meaningfully boosting productivity while humans retain decision authority. |
| Task automatability | claude-haiku-4-5-20251001 | 2/5 | While AI can assist with data analysis, risk modeling, and scenario testing, the task requires judgment about complex financial instruments, regulatory constraints, and institutional strategy that goes beyond pattern matching. Current systems cannot reliably replace the end-to-end expertise needed to advise financial institutions on investment or credit decisions at equal quality. |
| Task automatability | claude-sonnet-5 | 2/5 | This requires integrating regulatory knowledge, firm-specific risk appetite, negotiation, and judgment calls under uncertainty that current AI cannot autonomously perform end-to-end; AI can support analysis but not replace the advisory function. |
| Adoption barriers | claude-haiku-4-5-20251001 | 4/5 | Actuaries typically hold professional licensure (ASA, FSA, EA credentials), and financial institutions require sign-off from credentialed actuaries on risk assessments and product designs for regulatory compliance. These licensing and liability requirements create substantial barriers to full automation. |
| Adoption barriers | claude-sonnet-5 | 4/5 | Actuarial certification, regulatory filings, and fiduciary/liability requirements typically mandate credentialed human sign-off on risk assessments and pricing decisions. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 2/5 | Actuarial expertise commands high loaded wages ($150k+), and while AI can reduce some analytical labor, the cost of integrating specialized models, maintaining compliance, and ensuring accuracy for high-stakes decisions approaches human consultant costs rather than being significantly cheaper. |
| Cost vs. human wage | claude-sonnet-5 | 2/5 | AI can cheaply run models and simulations, but the overall advisory task still requires expensive expert review and liability-bearing sign-off, keeping blended cost closer to human-comparable. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 2/5 | AI tools exist for quantitative risk modeling and analytics, but no deployed product reliably performs the full scope of actuarial advisory—combining technical analysis with strategic recommendation and regulatory sign-off. Production systems handle components (pricing, reserve calculations) but not the integrated expert consultation this task demands. |
| Technical feasibility today | claude-sonnet-5 | 2/5 | Some analytics and modeling tools exist in production, but no deployed product provides holistic actuarial risk-and-returns advisory to institutions reliably without expert oversight. |
Manage credit and help price corporate security offerings.
26CI 25–28 · exposure 25 · augmentation 75 · importance 2.4/5 · click for rater detail
Manage credit and help price corporate security offerings.
26| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 2/5 | Despite operating in an information-rich sector, financial institutions have adopted AI cautiously for security pricing and credit management, prioritizing human expertise due to regulatory, reputational, and fiduciary risk. Production adoption remains limited compared to lower-stakes financial tasks. |
| Sector adoption velocity | claude-sonnet-5 | 3/5 | Financial services broadly is a fast-adopting sector for AI tools, but this specific niche task (corporate security pricing) sees more pilot-stage and augmentative tool use rather than widespread autonomous deployment. |
| Augmentation potential | claude-haiku-4-5-20251001 | 4/5 | AI substantially assists actuaries by automating data collection, running sensitivity analyses, and generating pricing scenarios, meaningfully improving productivity while the human actuary retains responsibility for final pricing decisions and risk judgment. |
| Augmentation potential | claude-sonnet-5 | 4/5 | AI-based analytics, risk models, and data aggregation tools meaningfully speed up credit analysis and pricing scenario generation, letting actuaries focus judgment on higher-level structuring and risk decisions. |
| Task automatability | claude-haiku-4-5-20251001 | 2/5 | Pricing corporate securities requires complex judgment about market conditions, risk factors, and client-specific context that AI cannot reliably execute end-to-end. While AI can assist with data analysis and some pricing model components, the task involves significant subjective decision-making and regulatory considerations that currently require human oversight. |
| Task automatability | claude-sonnet-5 | 2/5 | This involves complex judgment about credit risk, market conditions, and pricing strategy that requires synthesizing many qualitative and quantitative factors; current AI can assist with modeling components but cannot independently manage credit or price offerings end-to-end.' |
| Adoption barriers | claude-haiku-4-5-20251001 | 4/5 | Strong barriers exist: financial regulations (including disclosure requirements and liability rules) typically mandate that qualified actuaries and securities professionals sign off on pricing and credit decisions. Client relationships and institutional oversight also create friction against full automation. |
| Adoption barriers | claude-sonnet-5 | 4/5 | Corporate security pricing and credit risk management are subject to regulatory oversight, fiduciary responsibility, and often require credentialed actuaries or licensed professionals to sign off, creating strong liability and compliance barriers. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 2/5 | AI infrastructure and integration costs for financial modeling, compliance monitoring, and oversight remain substantial relative to the specialized human expertise required. The high stakes and regulatory demands mean supervision costs are significant. |
| Cost vs. human wage | claude-sonnet-5 | 2/5 | Sophisticated credit and pricing models require significant data infrastructure, human oversight, and validation, so all-in AI costs are not dramatically below skilled actuary/analyst labor costs for this specialized task. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 2/5 | No deployed AI product reliably performs complete credit assessment and security pricing independently. AI tools exist for data processing and model support, but they function as assistants within human-led workflows rather than autonomous performers of the full task. |
| Technical feasibility today | claude-sonnet-5 | 2/5 | While AI-driven credit risk models and pricing tools exist in production at some financial institutions, they typically serve as inputs to actuarial/analyst judgment rather than fully replacing the task, and error rates in novel security structures remain a concern. |
Provide advice to clients on a contract basis, working as a consultant.
16CI 6–25 · exposure 13 · augmentation 63 · importance 3.8/5 · click for rater detail
Provide advice to clients on a contract basis, working as a consultant.
16| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 2/5 | The actuarial profession has adopted AI for analytics and modeling support, but end-to-end consulting automation is not prevalent in production. Adoption remains at the pilot and assistive stage rather than replacement-level deployment. |
| Sector adoption velocity | claude-sonnet-5 | 2/5 | Actuarial and insurance consulting is a conservative, highly regulated professional services niche where AI pilots exist but full agent-based consulting substitution is rare. |
| Augmentation potential | claude-haiku-4-5-20251001 | 3/5 | AI systems can meaningfully assist with data preparation, scenario modeling, and pattern spotting in large datasets, allowing human actuaries to focus on judgment and client relationship. However, augmentation is limited to analytical components rather than the strategic advisory core. |
| Augmentation potential | claude-sonnet-5 | 4/5 | AI can strongly assist actuarial consultants with data analysis, report drafting, scenario modeling, and research, meaningfully boosting productivity while the human remains the responsible advisor. |
| Task automatability | claude-haiku-4-5-20251001 | 2/5 | Actuaries' consulting advice requires domain expertise, judgment about complex financial and risk scenarios, and personalized recommendations tailored to specific client contexts. While AI can assist with data analysis and calculations, the synthesis of technical knowledge with client-specific strategic judgment remains largely outside current automation scope. |
| Task automatability | claude-sonnet-5 | 1/5 | Client advisory consulting requires relationship-building, contextual judgment, negotiation, and accountability that current AI cannot replicate end-to-end; it is not a discrete, automatable workflow but an ongoing professional engagement. |
| Adoption barriers | claude-haiku-4-5-20251001 | 4/5 | Actuaries are credentialed professionals (ASA, FSA, EA credentials) whose work is often subject to regulatory scrutiny, professional liability standards, and client expectations of human expertise. Many jurisdictions and contexts require a qualified human actuary to sign off on advice. |
| Adoption barriers | claude-sonnet-5 | 5/5 | Actuarial advice often requires certification (e.g., Fellow of the Society of Actuaries), professional liability, and regulatory sign-off, making this a strongly protected, license-gated task. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 2/5 | Actuarial consulting commands high billable rates due to credential and expertise requirements; AI infrastructure and oversight costs do not yet undercut experienced human consultants, especially when liability and quality assurance are factored in. |
| Cost vs. human wage | claude-sonnet-5 | 2/5 | While AI can cheaply generate analysis or drafts, the actual consulting engagement still requires a credentialed human actuary for liability and trust reasons, so all-in cost savings are limited. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 2/5 | No deployed products reliably deliver full actuarial consulting advice end-to-end. AI tools can support analytics and modeling, but consulting requires validated expertise, accountability, and client trust that current systems cannot independently establish at production scale. |
| Technical feasibility today | claude-sonnet-5 | 1/5 | No deployed product independently acts as a contracted actuarial consultant advising clients; AI tools are used as internal aids by actuaries, not as autonomous consultants. |
Determine equitable basis for distributing surplus earnings under participating insurance and annuity contracts in mutual companies.
16CI 6–25 · exposure 13 · augmentation 75 · importance 3.3/5 · click for rater detail
Determine equitable basis for distributing surplus earnings under participating insurance and annuity contracts in mutual companies.
16| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 2/5 | Adoption of AI in actuarial work remains cautious, with firms using AI mostly for data processing and model calibration rather than autonomous decision-making. The regulatory environment and liability concerns slow production adoption of agents making distribution determinations. |
| Sector adoption velocity | claude-sonnet-5 | 2/5 | Insurance actuarial functions are conservative and heavily regulated, with AI adoption concentrated in narrower analytical support tasks rather than core equitable distribution determinations. |
| Augmentation potential | claude-haiku-4-5-20251001 | 4/5 | AI can substantially assist actuaries by rapidly modeling distribution scenarios, testing equity across cohorts, and flagging unintended consequences—significantly raising human productivity in design and analysis while the actuary retains final judgment. |
| Augmentation potential | claude-sonnet-5 | 4/5 | AI can meaningfully assist by running complex surplus allocation models, scenario analyses, and drafting documentation, significantly speeding up the analytical groundwork while the actuary retains final judgment. |
| Task automatability | claude-haiku-4-5-20251001 | 1/5 | This task requires nuanced judgment about fairness, stakeholder interests, and complex regulatory interpretation—domains where AI lacks reliable autonomy. No current system can independently determine 'equitable' distribution across competing policy-holder classes without expert human oversight and decision-making. |
| Task automatability | claude-sonnet-5 | 2/5 | This requires actuarial judgment, regulatory compliance, and equity considerations across policyholder classes that involve professional standards and legal responsibility; AI can support calculations but not independently determine the equitable basis end-to-end. |
| Adoption barriers | claude-haiku-4-5-20251001 | 4/5 | Actuaries are licensed professionals (ASA, FSA, EA credentials required), and regulatory frameworks (state insurance commissioners, IRS) mandate human actuarial sign-off on surplus distribution decisions. These legal and professional licensing barriers significantly restrict automation. |
| Adoption barriers | claude-sonnet-5 | 5/5 | Actuarial certification (e.g., Appointed Actuary opinions) and regulatory requirements mandate that a qualified, licensed actuary make and attest to such determinations, creating a hard legal barrier to full automation. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 2/5 | AI tools that assist with actuarial modeling have meaningful upfront setup costs and require integration with legacy insurance systems. The cost of inference and oversight is comparable to or exceeds hiring junior actuaries for data preparation and scenario testing. |
| Cost vs. human wage | claude-sonnet-5 | 2/5 | While AI can cheaply run scenario models and calculations, the human actuary's judgment, liability, and regulatory certification still dominate the cost structure, so total cost savings are limited. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 2/5 | While AI can assist with data aggregation and scenario modeling, no production system reliably makes autonomous decisions on surplus distribution. Deployed products may support analysis but require actuarial sign-off; the core judgment task remains human-centered. |
| Technical feasibility today | claude-sonnet-5 | 1/5 | No deployed product performs full surplus-distribution determination for mutual companies; this remains a specialized actuarial function requiring credentialed sign-off, not something current AI products handle in production. |
Negotiate terms and conditions of reinsurance with other companies.
13CI 5–20 · exposure 8 · augmentation 63 · importance 3.3/5 · click for rater detail
Negotiate terms and conditions of reinsurance with other companies.
13| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 1/5 | Reinsurance negotiation remains deeply human-centric in practice; adoption of AI for autonomous negotiation is minimal even in forward-looking firms. The sector is regulated and risk-averse, slowing experimentation with algorithmic deal-making. |
| Sector adoption velocity | claude-sonnet-5 | 2/5 | Insurance/reinsurance is a moderately digitized but conservative, heavily regulated sector where AI adoption in actual deal negotiation is nascent, mostly limited to data analysis support. |
| Augmentation potential | claude-haiku-4-5-20251001 | 3/5 | AI can assist by analyzing comparable deals, modeling scenarios, and summarizing contract clauses, improving actuaries' productivity in preparation and analysis. However, the core negotiation judgment remains human-dependent, limiting augmentation scope. |
| Augmentation potential | claude-sonnet-5 | 4/5 | AI can strongly assist by modeling risk scenarios, pricing analytics, and drafting term sheets, improving actuaries' preparation and negotiating leverage even though humans conduct the negotiation. |
| Task automatability | claude-haiku-4-5-20251001 | 1/5 | Negotiating reinsurance terms requires adversarial reasoning, understanding of counterparty interests, legal nuance, and real-time deal structure adjustment—tasks at which current AI systems cannot reliably perform end-to-end without human oversight. No AI system today can independently negotiate binding commercial agreements with novel partners and conditions. |
| Task automatability | claude-sonnet-5 | 2/5 | Negotiation involves real-time judgment, relationship management, and strategic trade-offs that current AI cannot reliably conduct end-to-end; at best AI can prep analysis or draft terms. |
| Adoption barriers | claude-haiku-4-5-20251001 | 4/5 | Reinsurance negotiation is governed by regulatory oversight, contractual liability, and professional licensing requirements; counterparties expect to deal with licensed actuaries and authorized representatives. Legal and fiduciary obligations create strong friction against full automation. |
| Adoption barriers | claude-sonnet-5 | 4/5 | Reinsurance contracts carry major financial and regulatory stakes requiring credentialed actuaries and authorized signatories, creating strong professional and liability barriers to automation. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 1/5 | The human actuaries and negotiators performing this task command high salaries (often $150k+), and current AI cannot reduce the end-to-end cost below that of employing skilled negotiators. AI analysis tools may assist but do not displace the core negotiation labor. |
| Cost vs. human wage | claude-sonnet-5 | 2/5 | Human actuaries/negotiators remain necessary for counterparty trust and deal-making, so AI cost savings are limited to prep work rather than the core negotiation itself. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 1/5 | No deployed product reliably negotiates reinsurance contracts autonomously; this remains a human-driven process despite AI tools for analysis. Deployed AI systems lack the legal authority, contextual judgment, and accountability required for independent negotiation in a licensed, heavily regulated domain. |
| Technical feasibility today | claude-sonnet-5 | 1/5 | No deployed product autonomously negotiates reinsurance contracts in production; this remains a human relationship-driven activity with AI only in supporting analytics. |
Testify before public agencies on proposed legislation affecting businesses.
1CI 0–3 · exposure 0 · augmentation 38 · importance 3.0/5 · click for rater detail
Testify before public agencies on proposed legislation affecting businesses.
1| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 1/5 | This is a task inherently requiring human presence, credibility, and accountability in a regulatory context; adoption of AI automation is not occurring because the task cannot and should not be delegated to machines. |
| Sector adoption velocity | claude-sonnet-5 | 2/5 | While actuarial and financial services broadly adopt AI tools, the specific act of public testimony sees essentially no AI penetration or displacement trend. |
| Augmentation potential | claude-haiku-4-5-20251001 | 2/5 | AI could assist an actuary in preparing testimony by drafting talking points or organizing evidence, but the live testimony itself must be delivered by the human expert; augmentation is limited to pre-testimony support. |
| Augmentation potential | claude-sonnet-5 | 3/5 | AI can meaningfully help actuaries research legislation, draft talking points, and analyze impacts beforehand, though it does not participate in the testimony itself. |
| Task automatability | claude-haiku-4-5-20251001 | 1/5 | Testifying before public agencies requires real-time oral advocacy, persuasive judgment, and responsiveness to questions—capabilities that current AI systems cannot reliably execute in a live, high-stakes legislative context. No meaningful part of this task can be automated end-to-end today. |
| Task automatability | claude-sonnet-5 | 1/5 | Live testimony before a public agency requires real-time human presence, judgment under questioning, and personal accountability that current AI cannot perform end-to-end. |
| Adoption barriers | claude-haiku-4-5-20251001 | 5/5 | Testifying before public agencies is a high-stakes activity where human professional judgment, accountability, and legal standing are non-negotiable. Regulatory and reputational barriers effectively prohibit AI substitution. |
| Adoption barriers | claude-sonnet-5 | 5/5 | Testimony typically requires a credentialed, accountable individual to appear, answer questions, and take professional/legal responsibility for statements, making this a hard human-only barrier. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 1/5 | The cost of AI inference and integration for this task would not be lower than hiring a qualified actuary, since the task cannot be reliably automated and human expertise remains legally and reputationally required. |
| Cost vs. human wage | claude-sonnet-5 | 1/5 | There is no AI substitute delivering this output, so cost comparison favors the human by default; any AI use is limited to prep support, not replacement of the act itself. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 1/5 | No deployed AI product today can autonomously testify before public agencies or represent an organization in legislative proceedings. This remains a human professional activity with no production-grade AI substitute. |
| Technical feasibility today | claude-sonnet-5 | 1/5 | No deployed product testifies before legislative or regulatory bodies on behalf of a professional; this remains entirely a human function. |
Testify in court as expert witness or to provide legal evidence on matters such as the value of potential lifetime earnings of a person disabled or killed in an accident.
0CI 0–0 · exposure 0 · augmentation 63 · importance 2.7/5 · click for rater detail
Testify in court as expert witness or to provide legal evidence on matters such as the value of potential lifetime earnings of a person disabled or killed in an accident.
0| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 1/5 | Legal testimony is governed by strict procedural and evidentiary rules that prohibit non-human witnesses. Adoption velocity is effectively zero because the task cannot be legally performed by AI in any jurisdiction. |
| Sector adoption velocity | claude-sonnet-5 | 1/5 | Legal and courtroom procedures are highly conservative and slow to adopt any technology that changes who may testify, with essentially no movement toward AI witnesses. |
| Augmentation potential | claude-haiku-4-5-20251001 | 3/5 | AI can assist an actuary in preparing testimony by organizing data, generating visualizations, calculating lifetime earnings scenarios, and drafting supporting arguments—useful productivity gains—but the actuary must deliver and defend all expert opinions in court. |
| Augmentation potential | claude-sonnet-5 | 4/5 | AI can meaningfully assist actuaries in preparing calculations, drafting reports, researching precedent, and organizing exhibits used to support their testimony, even though it cannot testify itself. |
| Task automatability | claude-haiku-4-5-20251001 | 1/5 | This task requires legal testimony, credibility establishment, cross-examination resilience, and real-time judgment calls in an adversarial setting—none of which current AI can perform end-to-end. AI cannot testify under oath or take legal responsibility for expert opinions. |
| Task automatability | claude-sonnet-5 | 1/5 | Live courtroom testimony requires a credentialed human to physically appear, respond to cross-examination, and be sworn under oath; no AI system can perform this end-to-end today. |
| Adoption barriers | claude-haiku-4-5-20251001 | 5/5 | Legal systems explicitly require a qualified human expert to testify under oath and be subject to cross-examination and perjury liability. Courts will not accept AI as a witness, and professional liability and evidentiary rules create hard barriers to substitution. |
| Adoption barriers | claude-sonnet-5 | 5/5 | Courts require a qualified, credentialed human expert witness who can be cross-examined and held legally accountable, making this a hard legal/licensing barrier to automation. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 1/5 | An actuary's testimony is billed at high expert-witness rates (often $300–$500+ per hour in court). AI cannot replace this revenue stream, and oversight by a human actuary would still be required, making total cost higher than human-only. |
| Cost vs. human wage | claude-sonnet-5 | 1/5 | AI cannot substitute for the sworn human testimony itself, so there is no viable AI cost basis to compare against the actuary's fee for this specific task. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 1/5 | No deployed product can serve as an expert witness in court; legal systems universally require a human professional to testify, be sworn in, and face cross-examination. AI assistance in preparing testimony exists, but AI cannot perform the task itself. |
| Technical feasibility today | claude-sonnet-5 | 1/5 | No deployed product provides in-court expert testimony; this remains squarely a human legal/professional function with no automation product on the market. |
Related occupations — Computer & Mathematical
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.