Pith. sign in

REVIEW 3 major objections 6 minor 88 references

Frontier AI exposure is sharply uneven across 141 countries, so U.S.- or Europe-calibrated policy will not travel.

Reviewed by Pith at T0; open to challenge. T0 means a machine referee read the full paper against a public rubric. the ladder, T0–T4 →

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

T0 review · grok-4.5

2026-07-12 14:31 UTC pith:IAIKG66V

load-bearing objection Solid, usable 141-country exposure map with a real gender pattern and a remittance channel; the construction is standard, the soft spots are already flagged, and it deserves referee time. the 3 major comments →

arxiv 2607.05404 v1 pith:IAIKG66V submitted 2026-06-08 cs.CY cs.AI

The Jagged Global Economy: Frontier AI Unevenly Exposes National Economies

Arul Murugan , Tom\'as Aguirre , Abhishek Nagaraj , Rishi Bommasani This is my paper
classification cs.CY cs.AI
keywords national AI exposurefrontier AIlabor marketsoccupational compositiongender gapremittancesAI adoptioncross-country comparison
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

This paper builds a national AI exposure score for 141 countries by weighting occupation-level exposure by each country's actual employment mix. It finds that high-income economies and Europe & Central Asia are far more exposed than low-income economies and Sub-Saharan Africa, that women are more exposed than men in 91% of countries because they are concentrated in white-collar and sales work, and that the scores predict real national AI adoption reported by major AI companies. It also identifies an indirect channel: remittance-dependent countries inherit exposure from the foreign labor markets that fund them. The authors argue that this cross-country variation is large enough that labor and AI policy written for the United States or Europe cannot safely be generalized.

Core claim

National AI exposure, defined as the employment-share-weighted average of occupation-level frontier-AI exposure, varies by a factor of about 2.6 across 141 countries. High-income countries and Europe & Central Asia are substantially more exposed than low-income countries and Sub-Saharan Africa; women are more exposed than men in 91% of countries; the scores predict Anthropic, Microsoft, and OpenAI adoption statistics; and remittance dependence can raise a country's effective exposure well above its domestic score.

What carries the argument

National AI exposure (equation 1): the employment-share-weighted average of nation-agnostic occupation exposure scores applied to each country's ISCO-08 employment structure. The object turns jagged task-level AI capability into a comparable national labor-composition index and supports both direct rankings and remittance-adjusted indirect exposure.

Load-bearing premise

The same occupation is treated as equally exposed in every country, so national differences come only from how many people work in each occupation, not from how the job is actually done in that country.

What would settle it

If, for the same ISCO-08 2-digit occupations, measured task content or observed AI usage differs systematically between high- and low-income countries enough to reverse the national ranking or the remittance-adjusted ordering for major remittance corridors, the central claim that domestic employment composition alone drives large, policy-relevant exposure gaps would fail.

Watch this falsifier — get emailed when new claim-graph text bears on it.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit.

Referee Report

3 major / 6 minor

Summary. The paper constructs a national AI exposure index for 141 countries as the employment-share-weighted average of ISCO-08 2-digit occupational exposure scores (Gmyrek et al.) using ILO employment data (Eq. 1). It reports large cross-country variation (Luxembourg ~2.6× Burundi), higher exposure in high-income economies and Europe & Central Asia than in Sub-Saharan Africa, a gender gap (women more exposed in 91% of countries, driven by white-collar and sales concentration), predictive correlations with Anthropic/Microsoft/OpenAI national adoption statistics, and an indirect remittance channel that raises measured exposure for remittance-dependent countries such as Tajikistan. The central policy claim is that this variation is large enough that U.S./Europe-calibrated AI labor policy will not generalize.

Significance. If the descriptive patterns hold, the paper fills a clear gap: most frontier-AI labor evidence is U.S./high-income-centric, while national occupational composition differs sharply. Strengths include transparent construction of n_i, public-facing country coverage (Table 4), robustness of national rankings to alternative occupational exposure indices (Appendix B.1; Spearman ρ ≈ 0.99), honest reporting that white-collar share alone yields R² = 0.91 (Figure 3) and that exposure adds little once white-collar share and log GNI enter adoption regressions (Table 6), and a novel remittance-mediated exposure channel with concrete bilateral examples. These are useful inputs for comparative AI policy and for prioritizing where richer microdata are most needed.

major comments (3)
  1. [§4, Table 6] §4 and Table 6: Exposure alone predicts Anthropic/OpenAI/Microsoft adoption (R² 0.61–0.81), but once white-collar share and log GNI are included the exposure coefficient is near zero and full-model R² is essentially unchanged. The manuscript should state more clearly what this implies for the metric’s incremental content: is n_i mainly a convenient summary of white-collar composition, or does it still justify separate policy weight? Without that discussion, the validation section overstates the distinctiveness of the exposure object relative to standard development covariates.
  2. [§5, Figure 5] §5 / Figure 5: Remittance-accounted exposure is load-bearing for the “new mechanism” claim, but the paper never writes the formula. Readers cannot tell whether remittance-accounted n is a GDP-share blend of domestic n_i and source-country n_j, a remittance-inflow-weighted average only, or something else; nor how 2021 bilateral shares are combined with 2024 remittance/GDP totals. An explicit equation parallel to Eq. (1), plus a one-line sensitivity to bilateral-year mismatch, is needed before the Tajikistan/Honduras-type rankings can be treated as comparable to direct exposure.
  3. [Abstract, §6] Abstract and §6: The leap from “exposure composition varies” to “policy responses calibrated to U.S. or European labor markets will not generalize” is stronger than the evidence. The paper correctly defines exposure as technical feasibility, not employment or wage impact (§1, §4), and does not show that optimal policy instruments differ by exposure level. Either soften the claim to “composition-based exposure differs enough that U.S./EU benchmarks are poor priors for many countries,” or add a short argument linking exposure gaps to concrete policy margins (e.g., training, social insurance, remittance risk).
minor comments (6)
  1. [Abstract vs §5] Abstract states “37 percent of Tajikistan GDP is Russian remittance” while §5 reports 47.9% remittances/GDP with Russia 79% of bilateral inflows (~37.8% product). Align wording so the abstract does not look like a different statistic.
  2. [Limitations, Figure 1] Limitations correctly flag missing China, Canada, and Saudi Arabia. A short note on how their omission affects regional averages (especially “Europe & Central Asia vs Sub-Saharan Africa”) would help readers bound the map in Figure 1.
  3. [§1 footnote / related work] Concurrent Gmyrek et al. [2026] is cited as similar methodology without published per-country scores. Clarify overlap vs contribution (your published 141-country panel, gender split, remittance channel, multi-provider adoption validation) in one paragraph so novelty is unambiguous.
  4. [§3.1, Figure 3] Figure 2 and white-collar definition: sales (ISCO 52) is treated as blue-collar in the main split but reclassified in a robustness note that lifts R² to 0.96. Flag the main-text definition once in the figure caption or §3.1 so readers do not reverse-engineer it from Appendix A.
  5. [Appendix B.1] Appendix B.1 crosswalk (O*NET/SOC → ISCO) is many-to-many; the unweighted-mean choice is stated, but a one-sentence note that part-share weighting barely moves correlations (already in footnote) should appear in the main robustness paragraph for non-appendix readers.
  6. [Figure 4, Table 6, A.5] Minor consistency: Microsoft complete-case n is 107 in Figure 4 caption and 106 in Table 6 / A.5. Reconcile.

Circularity Check

0 steps flagged

No significant circularity: national exposure is an external-data weighted average (ILO employment × independent Gmyrek occupational scores) validated against separate adoption and remittance sources.

full rationale

The core object is defined by Eq. (1) as the employment-share-weighted average of nation-agnostic occupational exposure scores taken from Gmyrek et al. (ILO Working Paper) and ILOSTAT 2-digit employment counts. Neither the adoption statistics (Anthropic Economic Index, OpenAI Signals ranks, Microsoft AI Diffusion rates) nor the KNOMAD remittance matrix enter the construction of n_i; they are used only for post-hoc correlational validation and a transparent second-order reweighting. Alternative occupational indices (Eloundou; Hosseini Maasoum & Lichtinger) produce near-identical national rankings (Spearman ρ ≈ 0.99), confirming that the ranking is not an artifact of a single fitted source. White-collar share and log GNI are shown to be strong predictors of n_i, but the paper does not claim that n_i is derived from them; the regressions are descriptive. No self-definitional loop, no fitted parameter renamed as an independent prediction, and no load-bearing uniqueness theorem imported from overlapping authors. The nation-invariant occupation scores are an acknowledged modeling assumption (Limitations), not a circularity. Score 0 is therefore appropriate.

Axiom & Free-Parameter Ledger

0 free parameters · 5 axioms · 2 invented entities

The central claims rest on a standard task-based exposure framework plus three modeling choices: nation-invariant occupation exposure, employment-share aggregation to a national index, and remittance-weighted blending of partner-country exposure. No free parameters are fitted to produce the main national scores; occupational exposures are taken from prior work. Invented constructs are measurement objects (national exposure and remittance-accounted exposure), not physical entities. The ledger is therefore light on free parameters and heavier on domain assumptions about cross-country occupational equivalence and remittance as an income-exposure channel.

axioms (5)
  • domain assumption Task/occupation exposure to frontier AI can be summarized by a scalar occupation score that is comparable across countries when occupations are coded in ISCO-08.
    Invoked throughout §2.2–§3; national metric uses Gmyrek et al. scores as nation-agnostic inputs to Eq. 1.
  • ad hoc to paper National AI exposure is well-defined as the employment-share-weighted average of occupation exposures (Eq. 1).
    This aggregation choice is the paper's operational definition of the national object; alternative weightings (wage-bill, hours, residual-task expertise) are not used for the main claims.
  • domain assumption Within a 2-digit ISCO category, cross-occupation and cross-country task differences can be ignored for ranking national exposure.
    Stated in §3 and Limitations; authors note concurrent work tries nation-sensitive occupation exposure but they do not.
  • domain assumption Bilateral remittance shares from the 2021 KNOMAD matrix, combined with later remittance-to-GDP totals, adequately represent current income dependence for remittance-accounted exposure.
    §5 and §A.4 freeze bilateral source shares at 2021 while using newer WDI totals; no re-estimation of corridors.
  • domain assumption Observed chatbot/generative-AI usage rates are valid external checks that national exposure captures economically relevant exposure, even though exposure is not itself adoption or displacement.
    §4 validation design; authors cite emerging evidence that exposure predicts adoption and some employment effects.
invented entities (2)
  • National AI exposure metric n_i independent evidence
    purpose: Summarize how strongly a country's current occupational mix aligns with frontier-AI-exposed tasks for cross-country comparison.
    Defined in Eq. 1 as employment-weighted average of occupation exposures; measurement construct rather than a new causal force.
  • Remittance-accounted national AI exposure no independent evidence
    purpose: Capture indirect exposure when national income depends on remittances from more AI-exposed host economies.
    Introduced in §5; blends domestic exposure with source-country exposures weighted by remittance shares. Independent handle exists via remittance-to-GDP and bilateral matrices, but the blended index itself is paper-defined.

pith-pipeline@v1.1.0-grok45 · 25200 in / 3670 out tokens · 33809 ms · 2026-07-12T14:31:01.898222+00:00 · methodology

0 comments
read the original abstract

Frontier AI's labor-market effects matter to workers, firms, and policymakers, but current evidence generally comes from a handful of high-income economies. The capabilities of frontier AI are jagged across work tasks and national economies diverge in how they allocate human labor. We introduce a national AI exposure metric that combines occupation-level exposure scores and international employment data for 141 countries. We find that high income countries are substantially more exposed than low income countries and that Europe and Central Asia are 50 percent more exposed than Sub-Saharan Africa. We also find a gender gap: women are more exposed than men in 91 percent of countries, driven by their concentration in white-collar and sales occupations. The exceptions are countries where women's employment remains concentrated in agriculture and household enterprises. We validate our national AI exposure estimates by showing they predict national AI adoption statistics published by Anthropic, Microsoft, and OpenAI. Beyond direct exposure, we identify a new mechanism for indirect exposure due to cross-country income dependencies. Some nations such as Tajikistan depend heavily on foreign workers remitting money back to their home countries: Tajikistan's direct exposure to frontier AI is below-average but because 37 percent of Tajikistan GDP is Russian remittance and Russia is very exposed, Tajikistan's remittance-accounted exposure becomes above-average. Our research shows that national variation in exposure is large enough that policy responses calibrated to U.S. or European labor markets will not generalize.

Figures

Figures reproduced from arXiv: 2607.05404 by Abhishek Nagaraj, Arul Murugan, Rishi Bommasani, Tom\'as Aguirre.

Figure 1
Figure 1. Figure 1: National AI exposure varies substantially across countries. Darker countries are more exposed to frontier AI. Countries without 2-digit ISCO-08 employment data are shown in gray. We introduce a national AI exposure metric to capture this heterogeneity. Our approach combines occupational exposure estimates from the economics of frontier AI literature with internationally comparable statistics on how countri… view at source ↗
Figure 2
Figure 2. Figure 2: Illustrative cross-country differences in labor composition [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: White-collar employment share predicts national AI exposure. The correlation between national AI exposure, white collar work, and national income manifests via other channels. Cross-country surveys find substantial heterogeneity in public opinion on AI on matters spanning trust in the technology, confidence in its benefits, concerns about its risks, and desire for regulation. For example, respondents in ad… view at source ↗
Figure 4
Figure 4. Figure 4: National AI exposure predicts national AI adoption. (a) Anthropic Claude usage per 100,000 working-age people, using the Anthropic Economic Index Claude.ai country release for February 5–12, 2026, and plotted on a log scale. (b) OpenAI Signals country-rank percentile for calendar-year 2025, based on OpenAI’s population-normalized ranking of countries by per-capita share of sampled consumer ChatGPT messages… view at source ↗
Figure 5
Figure 5. Figure 5: Remittance-accounted vs. direct national AI exposure. Countries shown have remittance of at least 10% of national GDP. We observe that while the concern is often expressed as AI suppressing wages, the underlying concern is that individuals will not be able to generate the income required to sustain their lives. In most countries, this can be restated in terms of wages, because individual income and individ… view at source ↗
Figure 6
Figure 6. Figure 6: National exposure estimates are robust to alternative occupational exposure measures. The top row compares ISCO-08 2-digit occupational exposure scores. The bottom row compares national exposure scores after reweighting each occupational index by country-level occupational employment. The alternative indices differ in their exact task definitions and source occupational taxonomies, but they produce very si… view at source ↗
Figure 7
Figure 7. Figure 7: Distribution of national AI exposure estimates. The side panels list the ten least- and most-exposed countries subject to the labor-force threshold used for visual readability; the histogram summarizes the full measured sample. 18 [PITH_FULL_IMAGE:figures/full_fig_p018_7.png] view at source ↗
Figure 8
Figure 8. Figure 8: visualizes the country-level gender exposure gap discussed in the main text. The x-axis reports the female-minus-male exposure gap as a percentage of total national exposure, so positive values indicate that female exposure exceeds male exposure. The pattern is broad rather than driven by a handful of countries: women are more exposed than men in most measured countries, while the largest negative gaps app… view at source ↗

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

88 extracted references · 2 canonical work pages

  1. [1]

    Science , volume =

    Tyna Eloundou and Sam Manning and Pamela Mishkin and Daniel Rock , title =. Science , volume =. 2024 , doi =. https://www.science.org/doi/pdf/10.1126/science.adj0998 , abstract =

    doi:10.1126/science.adj0998 2024

  2. [2]

    Artificial Intelligence and the Great Divergence , year =

  3. [3]

    and Karger, Ezra , title =

    Murphy, Connacher and Rosenberg, Josh and Canedy, Jordan and Jacobs, Zach and Flechner, Nadja and Britt, Rhiannon and Pan, Alexa and Rogers-Smith, Charlie and Mayland, Dan and Buffington, Cathy and Kučinskas, Simas and Coston, Amanda and Kerner, Hannah and Pierson, Emma and Rabbany, Reihaneh and Salganik, Matthew and Seamans, Robert and Su, Yu and Tramèr,...

  4. [4]

    Tetlock , title =

    Ezra Karger and Otto Kuusela and Jason Abaluck and Kevin Bryan and Basil Halperin and Todd Jones and Connacher Murphy and Phil Trammell and Matt Reynolds and Dan Mayland and Ria Viswanathan and Ananaya Mittal and Rebecca Ceppas de Castro and Josh Rosenberg and Philip E. Tetlock , title =. 2026 , month = mar, url =

    2026

  5. [5]

    2026 , eprint=

    How Well Does Agent Development Reflect Real-World Work? , author=. 2026 , eprint=

    2026

  6. [6]

    Task-Completion Time Horizons of Frontier AI Models , author =

  7. [7]

    Advances in Neural Information Processing Systems , year =

    Measuring AI Ability to Complete Long Software Tasks , author =. Advances in Neural Information Processing Systems , year =

  8. [8]

    arXiv preprint arXiv:2510.04374 , year =

    GDPval: Evaluating AI Model Performance on Real-World Economically Valuable Tasks , author =. arXiv preprint arXiv:2510.04374 , year =

    Pith/arXiv arXiv

  9. [9]

    arXiv preprint arXiv:2510.26787 , year =

    Remote Labor Index: Measuring AI Automation of Remote Work , author =. arXiv preprint arXiv:2510.26787 , year =

    arXiv

  10. [10]

    arXiv preprint arXiv:2509.25721 , year =

    The AI Productivity Index (APEX) , author =. arXiv preprint arXiv:2509.25721 , year =

    arXiv

  11. [11]

    2025 , month = apr, day =

    AI as Normal Technology: An alternative to the vision of AI as a potential superintelligence , author =. 2025 , month = apr, day =

    2025

  12. [12]

    Holistic Agent Leaderboard: The Missing Infrastructure for AI Agent Evaluation , author =

  13. [13]

    arXiv preprint arXiv:2602.16666 , year =

    Towards a Science of AI Agent Reliability , author =. arXiv preprint arXiv:2602.16666 , year =

    Pith/arXiv arXiv

  14. [14]

    2024 , month = may, doi =

    Daron Acemoglu , title =. 2024 , month = may, doi =

    2024

  15. [15]

    2026 , month = jan, url =

    Dario Amodei , title =. 2026 , month = jan, url =

    2026

  16. [16]

    2025 , month = oct, day =

    Bharat Chandar , title =. 2025 , month = oct, day =

    2025

  17. [17]

    2026 , month = jan, day =

    Alex Imas , title =. 2026 , month = jan, day =

    2026

  18. [18]

    Ipsos Predictions Survey 2025: Positivity about how this year has gone highest since before the pandemic , year =

    2025

  19. [19]

    2025 , url =

    Human Development Report 2025: A Matter of Choice: People and Possibilities in the Age of AI , institution =. 2025 , url =

    2025

  20. [20]

    2025 , url =

    AI Index Report 2025: Public Opinion , institution =. 2025 , url =

    2025

  21. [21]

    and Levy, Frank and Murnane, Richard J

    Autor, David H. and Levy, Frank and Murnane, Richard J. , title =. The Quarterly Journal of Economics , volume =

  22. [22]

    Handbook of Labor Economics , editor =

    Acemoglu, Daron and Autor, David , title =. Handbook of Labor Economics , editor =. 2011 , doi =

    2011

  23. [23]

    Journal of Economic Perspectives , volume =

    Acemoglu, Daron and Restrepo, Pascual , title =. Journal of Economic Perspectives , volume =. 2019 , doi =

    2019

  24. [24]

    AEA Papers and Proceedings , volume =

    Brynjolfsson, Erik and Mitchell, Tom and Rock, Daniel , title =. AEA Papers and Proceedings , volume =. 2018 , doi =

    2018

  25. [25]

    and Raj, Manav and Seamans, Robert , title =

    Felten, Edward W. and Raj, Manav and Seamans, Robert , title =. AEA Papers and Proceedings , volume =. 2018 , doi =

    2018

  26. [26]

    and Raj, Manav and Seamans, Robert , title =

    Felten, Edward W. and Raj, Manav and Seamans, Robert , title =. Strategic Management Journal , volume =. 2021 , doi =

    2021

  27. [27]

    2020 , doi =

    Webb, Michael , title =. 2020 , doi =

    2020

  28. [28]

    Measuring the Occupational Impact of AI: Tasks, Cognitive Abilities and AI Benchmarks , journal =

    Tolan, Song. Measuring the Occupational Impact of AI: Tasks, Cognitive Abilities and AI Benchmarks , journal =. 2021 , doi =

    2021

  29. [29]

    and Raj, Manav and Seamans, Robert , title =

    Felten, Edward W. and Raj, Manav and Seamans, Robert , title =. SSRN Electronic Journal , year =

  30. [30]

    and Cazzaniga, Mauro and Li, Longji , title =

    Pizzinelli, Carlo and Panton, Augustus and Tavares, Marina M. and Cazzaniga, Mauro and Li, Longji , title =

  31. [31]

    Generative AI and Jobs: A Refined Global Index of Occupational Exposure , institution =

    Gmyrek, Pawel and Berg, Janine and Kami. Generative AI and Jobs: A Refined Global Index of Occupational Exposure , institution =. 2025 , doi =

    2025

  32. [32]

    Mirror, Mirror on the Wall: Which Jobs Will AI Replace After All? A New Index of Occupational Exposure , institution =

    Ben. Mirror, Mirror on the Wall: Which Jobs Will AI Replace After All? A New Index of Occupational Exposure , institution =

  33. [33]

    2026 , url =

    Massenkoff, Maxim and McCrory, Peter , title =. 2026 , url =

    2026

  34. [34]

    and Immorlica, Nicole and Lucier, Brendan and Shahidi, Peyman , title =

    Demirer, Mert and Horton, John J. and Immorlica, Nicole and Lucier, Brendan and Shahidi, Peyman , title =. 2026 , doi =

    2026

  35. [35]

    and Goldfarb, Avi , title =

    Gans, Joshua S. and Goldfarb, Avi , title =. 2026 , doi =

    2026

  36. [36]

    http://www.nber.org/papers/w33941

    Autor, David and Thompson, Neil. Expertise. 2025. doi:10.3386/w33941 , URL = "http://www.nber.org/papers/w33941", abstract =

    work page doi:10.3386/w33941 2025

  37. [37]

    2025 , url =

    Artificial Intelligence Index Report 2025 , institution =. 2025 , url =

    2025

  38. [38]

    Ipsos AI Monitor 2025 , year =

    2025

  39. [39]

    2025 , month = jan, url =

    AI Opportunities Action Plan , howpublished =. 2025 , month = jan, url =

    2025

  40. [40]

    Canadian Sovereign AI Compute Strategy , year =

  41. [41]

    AI Continent Action Plan , year =

  42. [42]

    Introducing OpenAI for Countries , year =

  43. [43]

    2025 , url =

    Technology and Innovation Report 2025: Inclusive Artificial Intelligence for Development , institution =. 2025 , url =

    2025

  44. [44]

    Generative AI and Jobs: A Refined Global Index of Occupational Exposure , institution =

    Gmyrek, Pawe. Generative AI and Jobs: A Refined Global Index of Occupational Exposure , institution =. 2025 , doi =

    2025

  45. [45]

    and Pizzinelli, Carlo and Rockall, Emma J

    Cazzaniga, Mauro and Jaumotte, Florence and Li, Longji and Melina, Giovanni and Panton, Augustus J. and Pizzinelli, Carlo and Rockall, Emma J. and Tavares, Marina Mendes , title =. 2024 , month = jan, doi =

    2024

  46. [46]

    Workers' Exposure to

    Lewandowski, Piotr and Mado. Workers' Exposure to. 2025 , url =

    2025

  47. [47]

    Disruption without Dividend?: How the Digital Divide and Task Differences Split

    Gmyrek, Pawe. Disruption without Dividend?: How the Digital Divide and Task Differences Split. 2026 , doi =

    2026

  48. [48]

    , title =

    Brynjolfsson, Erik and Li, Danielle and Raymond, Lindsey R. , title =. The Quarterly Journal of Economics , volume =. 2025 , doi =

    2025

  49. [49]

    and Rajendran, Saran and Krayer, Lisa and Candelon, Fran

    Dell'Acqua, Fabrizio and McFowland III, Edward and Mollick, Ethan and Lifshitz, Hila and Kellogg, Katherine C. and Rajendran, Saran and Krayer, Lisa and Candelon, Fran. Navigating the Jagged Technological Frontier: Field Experimental Evidence of the Effects of Artificial Intelligence on Knowledge Worker Productivity and Quality , journal =. 2026 , doi =

    2026

  50. [50]

    ILOSTAT Database , year =

  51. [51]

    2024 , month = oct, doi =

    The Impact of Care Responsibilities on Women's Labour Force Participation , institution =. 2024 , month = oct, doi =

    2024

  52. [52]

    and Kahn, Lawrence M

    Blau, Francine D. and Kahn, Lawrence M. , title =. Journal of Economic Literature , volume =. 2017 , doi =

    2017

  53. [53]

    Proceedings of the National Academy of Sciences of the United States of America , volume =

    Humlum, Anders and Vestergaard, Emilie , title =. Proceedings of the National Academy of Sciences of the United States of America , volume =. 2025 , doi =

    2025

  54. [54]

    and Delecourt, Sol

    Otis, Nicholas G. and Delecourt, Sol. Global Evidence on Gender Gaps and Generative AI , institution =. 2024 , url =

    2024

  55. [55]

    2025 , month = apr, day =

    McClain, Colleen and Kennedy, Brian and Gottfried, Jeffrey and Anderson, Monica and Pasquini, Giancarlo , title =. 2025 , month = apr, day =

    2025

  56. [56]

    The Gen AI Gender Gap , institution =

    Aldasoro, I. The Gen AI Gender Gap , institution =. 2024 , month = jul, url =

    2024

  57. [57]

    2025 , month = oct, day =

    How People Around the World View AI , institution =. 2025 , month = oct, day =

    2025

  58. [58]

    and Macdade, A

    Gillespie, Nicole and Lockey, Steve and Ward, T. and Macdade, A. and Hassed, G. , title =. 2025 , doi =

    2025

  59. [59]

    2026 , month = mar, day =

    Huang, Saffron and Carter, Shan and Eaton, Jake and Pollack, Sarah and Callender III, Dexter and Makagiansar, Nikki and Gonzalez, Maria and Carr, Sylvie and Hong, Jerry and Handa, Kunal and McCain, Miles and Millar, Thomas and Julapalli, Mo and Yun, Grace and Alt, AJ and Larsson, Chelsea and Leibrock, Jane and Gallivan, Matt and Sumers, Theodore and Durmu...

    2026

  60. [60]

    http://www.nber.org/papers/w34705

    Manning, Sam J and Aguirre, Tomás. How Adaptable Are American Workers to AI-Induced Job Displacement?. 2026. doi:10.3386/w34705 , URL = "http://www.nber.org/papers/w34705", abstract =

    work page doi:10.3386/w34705 2026

  61. [61]

    arXiv preprint arXiv:2507.07935 , year =

    Working with AI: Measuring the Applicability of Generative AI to Occupations , author =. arXiv preprint arXiv:2507.07935 , year =. doi:10.48550/arXiv.2507.07935 , url =

    doi:10.48550/arxiv.2507.07935

  62. [62]

    2025 , month = sep, doi =

    How People Use ChatGPT , author =. 2025 , month = sep, doi =

    2025

  63. [63]

    arXiv preprint arXiv:2503.04761 , year =

    Which Economic Tasks Are Performed with AI? Evidence from Millions of Claude Conversations , author =. arXiv preprint arXiv:2503.04761 , year =. doi:10.48550/arXiv.2503.04761 , url =

    Pith/arXiv arXiv doi:10.48550/arxiv.2503.04761

  64. [64]

    2025 , month = nov, url =

    Canaries in the Coal Mine? Six Facts about the Recent Employment Effects of Artificial Intelligence , author =. 2025 , month = nov, url =

    2025

  65. [65]

    2025 , month = sep, day =

    Appel, Ruth and McCrory, Peter and Tamkin, Alex and Stern, Michael and McCain, Miles and Neylon, Tyler , title =. 2025 , month = sep, day =

    2025

  66. [66]

    2026 , month = feb, note =

    OpenAI Signals: Measuring AI Adoption, Protecting Privacy, and Empowering Decisions , author =. 2026 , month = feb, note =

    2026

  67. [67]

    , title =

    Tucker, Lee C. , title =. 2026 , month = apr, day =

    2026

  68. [68]

    2026 , month = mar, day =

    Imas, Alex and Shukla, Soumitra , title =. 2026 , month = mar, day =

    2026

  69. [69]

    2026 , howpublished =

    Bilateral Remittance Estimates using Migrant Stocks, Host Country Incomes, and Origin Country Incomes (. 2026 , howpublished =

    2026

  70. [70]

    2026 , howpublished =

    Remittance inflows (. 2026 , howpublished =

    2026

  71. [71]

    2022 , month = dec, howpublished =

    Ratha, Dilip and Plaza, Sonia and Kim, Eung Ju , title =. 2022 , month = dec, howpublished =

    2022

  72. [72]

    2023 , month = mar, number =

    Dick, Eva and Schraven, Benjamin and Ratha, Dilip and Kim, Eung Ju , title =. 2023 , month = mar, number =

    2023

  73. [73]

    4.9 million employed in finance and insurance sector , year =

  74. [74]

    2026 , month = feb, howpublished =

    Labour Force in Singapore 2025 , institution =. 2026 , month = feb, howpublished =

    2025

  75. [75]

    2024 , month = nov, day =

    Burundi: Supporting youth-at-risk through the creative sector (P505887) Project Information Document (PID) , institution =. 2024 , month = nov, day =

    2024

  76. [76]

    2024 , howpublished =

    2024 Fact Sheet on Women and Men in the Philippines: Work and Economic Participation , institution =. 2024 , howpublished =

    2024

  77. [77]

    2020 , month = jul, day =

    Jamaican Labour Market: Impact of COVID-19 , institution =. 2020 , month = jul, day =

    2020

  78. [78]

    2026 , howpublished =

    Labour Force Survey 2024--25 , institution =. 2026 , howpublished =

    2024

  79. [79]

    2023 , howpublished =

    Women and Men in India 2023: Participation in Economy , institution =. 2023 , howpublished =

    2023

  80. [80]

    O*NET 30.2 Data Dictionary , year =

Showing first 80 references.