DAIOE: how exposed is each job to AI?
DAIOE measures how exposed each occupation is to artificial intelligence, from data rather than expert guesswork. It tracks AI capability subdomains annually since 2010, capturing the potential applicability of AI capabilities to occupational content, not job-loss forecasts or adoption probabilities. It is openly published and citable (Zenodo, doi:10.5281/zenodo.21873968), versioned, and mapped across the US (SOC), international (ISCO) and Swedish (SSYK) classifications, so others can use it directly.
How exposed is your job?
Find your occupation.
Type an occupation to see its DAIOE exposure and where it sits among roughly 424 occupations. Generative AI by default; switch the sub-domain to compare. As featured in Die Zeit ↗
The whole landscape · generative AI, 2023
Every occupation, placed by its exposure.
Roughly 424 occupations, each a dot, placed left to right by how exposed they are to generative AI. Hover any dot to name it.
The exposed end is desk work. Writers, programmers, analysts, marketers and, yes, economists cluster on the right.
The other end is hands and bodies. Care, craft, construction, cleaning and farming sit on the left, where generative AI reaches least.
It cuts against intuition. The more schooling a job needs, the more exposed it tends to be. Exposure is not replacement, but the pattern is stark.
Source: DAIOE v2023 · ISCO-08. Look up your own job in the search above.
The named extremes · generative AI, 2023
Where generative AI reaches, and where it doesn't.
DAIOE's generative-AI exposure across roughly 424 occupations. Writers, marketers, programmers and, yes, economists sit at the very top; hands-on manual, craft and outdoor work sits at the bottom.
Source: DAIOE v2023 · ISCO-08 · higher score = more exposed. Explore every occupation in the Occupations Explorer.
The measure
Data-driven, not expert-guessed.
DAIOE scores each occupation's exposure to AI from data, and publishes those scores openly and with versions, mapped across the US (SOC), international (ISCO) and Swedish (SSYK) classifications, so others can join it straight onto their own data.
Use it in your own work
Open data & crosswalks.
- DAIOE scores on Zenodo (citable, all versions) · the canonical, versioned release; cite the vintage you use
- Construction pipeline (open source) · code, test suite, quick starts in Python and Stata
- Technical documentation (construction, vintages, reading the scores) · the release's public reference
- SSYK 2012 & SSYK 96 translation utility · crosswalk helper
Introduced and validated in the working paper “AI Unboxed and Jobs: A Novel Measure and Firm-Level Evidence from Three Countries”. See Research.
FAQ
What DAIOE is, and isn't.
Does “exposure” mean automation or job loss?
No. DAIOE measures how applicable AI’s advancing capabilities are to what an occupation does. Whether that changes jobs depends on adoption, complements and organisational choices.
Is it a forecast of AI adoption?
No. DAIOE captures potential applicability. If, when and how AI is applied in an occupation depends on the cost of acquiring, adapting and using the technology, the institutional setting (laws and regulations), and more.
What is different about DAIOE versus prior indices?
DAIOE is dynamic and grounded in measured AI benchmark progress: the score moves year by year as AI capabilities actually advance, rather than scoring each occupation once. It tracks separate capability subdomains over time and down-weights socially intensive work (see below).
How do I read the raw score values?
The index has no natural units: a value is meaningful only relative to other occupation-years. For cross-sectional standing, use within-year percentiles. One caution: v1.0.0’s rank columns can order occupations with identical values arbitrarily, so where ties could matter compute the tie-safe midrank instead (one line: 100 × average rank within year / count); from v1.1.0 it ships precomputed. Percentiles are ordinal, so they do not show how much exposure has grown; for level and growth, rescale as in the next question.
Can I express scores relative to the most exposed occupation?
Yes: divide by the highest value in the frozen 2010 to 2023 panel you use and multiply by 100 (score_rel_max = 100 x value / max). A score then reads as per cent of the most exposed occupation-year, and growth survives: an occupation at 31 in 2015 and 62 in 2023 doubled its exposure. Keep the 2010 to 2023 maximum as the denominator for later vintages too; values above 100 then mean exposure beyond the frozen-window peak. We document this transformation rather than shipping it, so the raw values remain the single citable version; details are in the technical documentation below.
Which occupations tend to score highest?
Cognitive, non-physical roles with lower social-interaction intensity, often white-collar, tend to rank higher.
Do the data say anything about employment outcomes?
Within Swedish firms, higher exposure predicts lower clerical support employment, an association absent in a pre-AI placebo period; skill ratios rise with exposure in Denmark and Portugal. There is no robust association with total employment. See the working paper for the full evidence and its limits.
How do you handle interpersonal-heavy work?
Through a down-weighting built into the measure. The abilities that generate the exposure score capture how applicable AI’s measured capabilities are to a job’s content, but they do not fully capture how interpersonal that content is, and the more a job consists of human interaction, the smaller the part of it AI’s capabilities apply to. DAIOE therefore scores social intensity separately: O*NET rates how important interpersonal skills are in each occupation and at what level the job requires them, and those ratings combine into a score that dials exposure down as it rises. Without such an adjustment, clergy rank as strongly exposed, yet the substance of their work is sustained human relationships, where the measured capabilities apply least. The adjustment tempers exposure rather than erasing it, and the paper shows the results do not hinge on how strongly it is applied.
What are the AI subdomains?
Capability areas across games, vision and language, each tracked from its own public benchmarks. In the frozen 2010–2023 index: abstract strategy games and real-time video games; image recognition, image comprehension and image generation; reading comprehension, language modelling, translation and speech recognition. They aggregate into the overall index, and language modelling and image generation also combine into the generative-AI sub-index. Later vintages add subdomains as credible benchmarks emerge, and each vintage documents its own set.
Can I compare across countries or industries?
Yes. Combine occupation-level DAIOE with local workforce structures (firms, industries, regions) to assess exposure distributions.
Is there a generative-AI version?
Yes. We provide a generative-AI sub-index that combines language modelling and image generation.
How do I cite DAIOE?
Cite the release and the paper, and name the object you used. The scores carry a DOI per version (this version: 10.5281/zenodo.21873968; all versions: 10.5281/zenodo.21873967). v1.0.0 contains two objects, so cite the DOI plus the object (frozen 2010–2023 index, or 2024 refresh) plus the filename you loaded; the repository's CITATION.cff carries the machine-readable form.
What is a vintage, and which should I use?
A vintage is one released version of the dataset, labelled by its coverage window; vintages are separate, citable objects and are not interchangeable. Use the frozen 2010–2023 index to replicate or compare against published work, and the 2024 refresh for current analysis. The frozen window is carried cell-identically in every later vintage: published values never change.
How do I standardise the scores the way the paper does?
The bundle ships the frozen 2010–2020 moments (standardisation_moments_v1.csv: mean and standard deviation per classification and column), so standardising is one line, z = (value − mean) / sd, and cannot be done on the wrong window.
How do I get going quickly?
The repository README opens with quick starts in Python and Stata: the full ranking of occupations with percentiles, top and bottom ten with titles, and one occupation's exposure path over time. A code–title lookup for the Swedish classifications ships in the bundle.
Will DAIOE be updated, and what happens to old values?
Yes, by annual vintages, with published values never changing: every release verifies cell by cell that the frozen window is identical, and all basket and membership changes take effect at annual chain points, the convention of chain-linked official statistics. A 2025 vintage is forthcoming as v1.1.0, adding agentic task execution and mathematical and scientific reasoning, with companion composites in standardised units.
See it live
DAIOE in the Monitor.
The Occupations Explorer, part of the AIEL Monitor, sets Swedish employment by occupation against DAIOE exposure levels, in yearly and monthly views.