aijobimpactcalculator.com
Menu
CalculatorOccupationsIndustriesWhat's growingMethodologyFAQ

Reference / Methodology / Last verified April 2026

Methodology: how this calculator works

The single most-important page on the site. Every claim on every other page traces back to this one. If you cannot defend a number from this page, the number does not appear elsewhere.

What this calculator does, and does not do

The calculator outputs three things for the matched occupation: an ILO 2025 generative AI exposure gradient (one of four bands), a task-level breakdown of the top five O*NET 30.2 tasks tagged Displaceable / Changing / Growing per the Brookings 2024 rubric, and a growth panel drawn from BLS Employment Projections 2024-2034 and WEF Future of Jobs Report 2025 skills demand. All values are static and pre-computed at build time. There are no runtime API calls.

The calculator does NOT predict whether you specifically will lose your job. It does not factor in your individual employer, geography, age, salary, or labour-market context. It does not give career, financial, or legal advice. It is a research synthesis at the occupation-family level.

The sources

The full bibliography with URLs and access dates is at /sources/. The summary classification:

  • Per-occupation, primary scoring: ILO 2025 refined Generative AI Occupational Exposure Index. ISCO-08 6-digit occupations across approximately 30,000 tasks, four exposure gradients per occupation. Cross-referenced with the OECD AI Working Group's 2024 sectoral and 2025 generative-AI work for triangulation.
  • Per-task, primary classification: Brookings Institution 2024, "Generative AI, the American worker, and the future of work". Task-level automation exposure across O*NET tasks for over 1,000 occupations. Used to tag each occupation's top five O*NET tasks.
  • Per-occupation, growth projection: US Bureau of Labor Statistics, Employment Projections 2024-2034. Per-SOC employment projections for approximately 800 detailed occupations through 2034.
  • Per-occupation, taxonomy and tasks: O*NET 30.2 (CC-BY 4.0). Source for the occupation list, top-five task statements per occupation, similar occupations, and skills profile.
  • Aggregate, skills demand: WEF Future of Jobs Report 2025. Top growing skills mapped to occupation skills profiles via O*NET data. Family-level, not per-occupation.
  • Aggregate context only: McKinsey Global Institute 2024, Goldman Sachs 2023. Cited for time-horizon framing on the homepage; not used as primary methodology to avoid consultant or disruptor optics.
  • Excluded: Frey-Osborne 2013 (pre-LLM, missed the cognitive-task disruption); Eloundou et al. (OpenAI-derived, disruptor-company optics).

The algorithm, in five steps

  1. Occupation match. User enters a job title. The calculator matches the input against the O*NET 30.2 occupation list and curated search aliases via a build-time fuzzy index over the 30 priority occupations published at launch. The match is rendered with the O*NET-SOC and ISCO-08 codes for verification.
  2. Exposure gradient. The matched O*NET-SOC code is mapped to ISCO-08 via the BLS-published SOC-to-ISCO crosswalk. Where the mapping is one-to-many, the calculator uses the dominant match (the first ISCO-08 code in the BLS crosswalk for that SOC code). The ILO 2025 refined index is queried for the gradient at that ISCO-08 code. The four-band output (Low / Moderate / High / Very High) is the result.
  3. Task breakdown. The top five tasks for the matched O*NET-SOC code are pulled from O*NET 30.2 by Importance score. Each task is classified using the Brookings 2024 rubric (technical feasibility plus contextual feasibility). Classifications are pre-computed and committed to source. They are not generated at runtime.
  4. Growth panel. The matched O*NET-SOC code is queried against the BLS 2024-2034 Employment Projections data. The projected percent change and absolute change are pulled. The top three growing skills relevant to the occupation are mapped from the WEF 2025 list to the O*NET skills profile by skill family.
  5. Output rendering. The three panels are rendered statically. The shareable PNG card per occupation is also pre-rendered at build time. No runtime API calls.

How the task tags are computed (Brookings 2024 rubric)

The Brookings 2024 task-level rubric distinguishes technical feasibility (whether current AI can perform the task) from contextual feasibility (whether an organisation can delegate the task to AI given workplace, regulatory, and accountability constraints). The calculator's tags are derived as follows:

  • Displaceable: technical feasibility AND contextual feasibility. The task is in scope for current generative AI and organisations are visibly delegating it.
  • Changing: technical feasibility but contextually constrained, OR the task is being augmented rather than replaced. The human stays in the loop; the work shifts.
  • Growing: the task is in the augmentation-prone category per Brookings 2024, OR the task is BLS-flagged as growing for the occupation.

Open questions and unresolved uncertainty

The pace of AI capability improvement is not known precisely. The ILO 2025 refined index is the latest published gradient at time of writing; it is updated annually by the ILO and refreshed here accordingly.

Labour-market response to exposure is uneven across employers, geographies, and time horizons. Brookings 2025 (No AI Jobs Apocalypse, For Now) reports aggregate-labour-market data does not yet show mass displacement; this could change. The calculator's exposure gradient does not predict displacement timing.

The calculator cannot resolve fine-grained variation within an occupation. A marketing manager at a SaaS company and a marketing manager at a regional non-profit are the same O*NET-SOC code and receive the same gradient. Within-occupation variance is real and sometimes larger than between-occupation variance. The site acknowledges this rather than invent precision the source data does not support.

Where the calculator's limitations are

The ILO 2025 gradient is one of four bands; it cannot resolve fine-grained differences within a band. Two occupations both in the High band may differ substantially in actual exposure.

The Brookings task-level rubric depends on O*NET task lists, which are updated quarterly but lag actual workplace tasks by months to years. AI-deployment patterns from 2025-2026 are not yet fully reflected in the public O*NET task statements.

The BLS projection cycle is decade-scale. The most recent published projection covers 2024-2034. Rapid 2024-2026 shifts may not yet be incorporated. The next projection cycle is expected late 2026 and will be incorporated when published.

How to argue with this calculator

Every reasonable challenge to the calculator's methodology is pre-empted at /how-to-argue-with-this/. Including: the OECD-to-O*NET crosswalk error, the Brookings dependency on OpenAI's task-completion data, the Frey-Osborne comparison, the BLS aggregate-versus-AI-specific concern, and the single-percentile-versus-band design choice.

Update cadence

O*NET pulls quarterly (next major in Q3 2026). ILO and OECD updates annually (ILO 2025 update in late 2025; ILO 2026 expected late 2026). BLS Employment Projections every two years (2024-2034 release was September 2025; 2026-2036 expected late 2026). WEF Future of Jobs biennially (2025 edition is current; 2027 edition expected early 2027).

The "Last verified" date in the page footer reflects the most-recent refresh. The current value is April 2026.

Author and editor

Built by Digital Signet. Corrections are welcome. The cluster sister site for the definitional framing of AI agents is whatisanaiagent.com.

Revision history

  • April 2026: initial publication.