Evidence (312 claims)
Search and filter individual claims pulled from the papers. Looking for a specific finding ("what's the effect on wages?"), you're in the right place. Want to compare whole outcome categories against each other instead? Use the Evidence Explorer.
The board below groups claims two ways: by broad theme (nine paper-level topics) and by outcome category (the 34 claim-level outcomes that the Explorer and Syntheses also use).
Browse by theme
Nine broad, paper-level topics. Click one to filter the claims below.
Adoption
20058 claims
Filter claims →
Productivity
17184 claims
Filter claims →
Governance
16099 claims
Filter claims →
Human-AI Collaboration
16034 claims
Filter claims →
Innovation
10501 claims
Filter claims →
Org Design
10496 claims
Filter claims →
Labor Markets
6444 claims
Filter claims →
Skills & Training
5385 claims
Filter claims →
Inequality
4148 claims
Filter claims →
Claims by outcome category
Counts by direction of finding. These are the same 34 outcome categories the Explorer compares and the Syntheses are written for. A linked row has a published synthesis.
| Outcome | Positive | Negative | Mixed | Null | Total |
|---|---|---|---|---|---|
| Other | 1820 | 479 | 278 | 1820 | 4588 |
| Organizational Efficiency | 2711 | 616 | 401 | 173 | 3922 |
| Governance & Regulation | 2075 | 886 | 459 | 246 | 3714 |
| Technology Adoption Rate | 1467 | 530 | 258 | 206 | 2488 |
| Decision Quality | 1281 | 496 | 289 | 152 | 2228 |
| Output Quality | 1227 | 447 | 207 | 138 | 2025 |
| AI Safety & Ethics | 634 | 754 | 207 | 83 | 1688 |
| Research Productivity | 826 | 241 | 114 | 422 | 1624 |
| Firm Productivity | 1052 | 154 | 163 | 66 | 1441 |
| Task Allocation | 685 | 211 | 331 | 99 | 1335 |
| Market Structure | 433 | 423 | 242 | 46 | 1150 |
| Innovation Output | 639 | 91 | 105 | 34 | 871 |
| Task Completion Time | 476 | 113 | 43 | 36 | 672 |
| Firm Revenue | 445 | 126 | 58 | 25 | 656 |
| Skill Acquisition | 364 | 119 | 109 | 34 | 626 |
| Consumer Welfare | 288 | 167 | 104 | 31 | 592 |
| Employment Level | 214 | 140 | 174 | 50 | 582 |
| Error Rate | 230 | 251 | 35 | 16 | 535 |
| Fiscal & Macroeconomic | 268 | 136 | 71 | 50 | 532 |
| Inequality Measures | 100 | 307 | 96 | 12 | 515 |
| Worker Satisfaction | 221 | 173 | 60 | 30 | 484 |
| Automation Exposure | 155 | 138 | 65 | 36 | 398 |
| Regulatory Compliance | 171 | 120 | 30 | 13 | 335 |
| Developer Productivity | 222 | 58 | 27 | 13 | 321 |
| Team Performance | 188 | 56 | 50 | 24 | 320 |
| Wages & Compensation | 146 | 104 | 46 | 16 | 312 |
| Training Effectiveness | 207 | 41 | 21 | 26 | 298 |
| Job Displacement | 23 | 153 | 52 | 4 | 232 |
| Hiring & Recruitment | 102 | 57 | 30 | 11 | 202 |
| Skill Obsolescence | 16 | 102 | 24 | 6 | 148 |
| Creative Output | 71 | 42 | 23 | 6 | 143 |
| Social Protection | 57 | 30 | 11 | 3 | 101 |
| Labor Share of Income | 29 | 42 | 24 | 2 | 97 |
| Worker Turnover | 43 | 29 | 6 | 4 | 82 |
| Industry | — | — | — | 1 | 1 |
High-skill cognitive tasks and complementary occupations gain earnings, while routine tasks and low-skill occupations face displacement and wage pressure.
Asserted in the supplied example contribution; no occupational employment or wage data are presented.
Highly skilled workers who complement AI are likely to receive a skill premium, whereas medium-skilled workers displaced by AI may experience stagnant or declining incomes if they do not successfully retrain.
The paper explains the claim through skill scarcity, complementarity, and displacement mechanisms and refers to micro-level studies and an IMF working paper, but reports no original estimates.
Digitalization-induced changes in labor-time supply—including flexibility, intensity, and hours—can affect earned income and contribute to distributional differences.
The model explicitly includes a supply-side labor-time modulation mechanism and evaluates its implications for earned income and inequality.
AI adoption in banking may shift labor demand toward skilled workers, potentially increasing wages for data- and AI-related roles while compressing demand and possibly wages for routine positions.
Economic interpretation and implications presented in the supplied text; no quantitative wage data or estimated effect sizes are reported.
Skill acquisition can increase worker productivity, but productivity gains do not automatically translate into higher wages.
Conceptual SPEW framework and simplified economic model linking skill enhancement, productivity, and potential wage gains; supported by qualitative observations and synthesis of existing programme evidence rather than a causal evaluation.
Automation-type AI depressed new-work creation, employment, and wages in lower-skilled occupations, whereas augmentation-type AI raised wages and generated new work in higher-skilled occupations.
Empirical work distinguishing automation-type from augmentation-type AI, as synthesized by the paper.
Uneven adoption of AI capabilities across multinational studios may change global skill and wage patterns by concentrating AI-intensive tasks in studios with stronger research and computational capacity.
Conceptual economic implication concerning geographic differences in AI adoption; the paper does not report wage or skill data.
This implies a non-monotonic relationship between labor share and wages.
Logical consequence of the theoretical result that the wage-maximizing labor share depends only on the capital-to-labor ratio (model-based derivation).
In a competitive economy with constant returns to scale, the wage-maximizing labor share depends only on the capital-to-labor ratio.
Analytic/theoretical proof within a competitive constant-returns-to-scale model (formal derivation in the paper).
The index is structured to reflect both direct substitution effects and indirect equilibrium adjustments, including wage compression, labor reallocation, and productivity spillovers.
Paper describes structural elements the index is designed to capture (direct substitution and various indirect adjustments); presented as theoretical design features without empirical estimates.
Task complexity mitigates the negative association between AI exposure and wages.
Heterogeneity/moderation analysis showing that occupations (or tasks) with higher complexity exhibit a smaller (i.e., mitigated) negative wage effect from AI exposure.
The effect of the model's structural parameters on key variables such as wages is quantified.
Results from high-throughput model simulations / quantitative sensitivity analysis of structural parameters reported for wages.
Wage increases are concentrated in technical and AI-complementary roles, which are predominantly male-dominated.
Analysis of wage trends by occupational group and gender from the 2015–2025 longitudinal dataset showing larger wage increases for technical/AI-complementary occupations and a gender composition skewed toward men in those occupations.
The study examines direct effects of AI and automation on employment levels, (re)training and occupational mobility, wage dynamics, collective bargaining, and the emergence and resolution of collective labor disputes and strikes.
Scope statement of the paper; supported by literature review (OECD, ILO) and legal/institutional analysis; no specific empirical sample or experiment reported in the summary.
Adoption of AI affects wages.
Mention and discussion in the paper (index entry referencing page 89) — theoretical discussion and literature synthesis.
Using these averages, we report on likely tradeoffs between salaries and AI exposure across interest categories, O*NET Job Zones, and job fields.
Analyses performed on averaged exposure projections combined with occupation-level salary, O*NET Job Zone, interest category, and job field data (descriptive cross-tabulations / comparative statistics as reported).
We measure the change in the skill premium using a difference-in-differences design on freelance websites worldwide.
Statement of empirical method: difference-in-differences design applied to data from freelance platforms with global coverage; no sample size provided in the abstract.
Instrumental-variable estimates using lagged AI diffusion produce similar patterns (attenuation of overeducation penalty and slight lowering of undereducation premium), although results should be interpreted with caution.
IV estimation using lagged AI diffusion as an instrument in models applied to CLDS data; IV results reported to be qualitatively similar to OLS/fixed-effects estimates but noted as requiring cautious interpretation.
Other strategic factors (differentiation of work, digital reputation, adaptability) continue to influence illustrators' financial sustainability despite AI's effect.
Author conclusion/interpretation in the discussion, inferred from the relatively low R² and domain knowledge; these moderators/alternative determinants are asserted rather than estimated in the reported regression.
AI explains a relatively small share of income variation among illustrators (model R² = 7.4%), so its contribution to income variation is limited.
Reported model fit statistic from the above simple linear regression (R² = 7.4%) on the sample of 385 illustrators.
Robustness checks across the capital share, shock persistence, and the utility specification show that only an empirically implausible labor–AI elasticity reverses the wage and fertility signs.
Sensitivity/robustness analysis of model results by varying parameters (capital share, shock persistence, utility functional form) and the labor–AI elasticity, reporting conditions under which sign flips occur.
Industry-wise, sectors with higher levels of digitalization (e.g., mining, finance, energy) show stronger income effects, while traditional sectors (e.g., agriculture, public services) show limited impact.
Industry-level heterogeneity analysis in the two-way fixed effects panel using provincial data (2011–2021), reporting larger estimated effects for high-digital sectors and small or null effects for traditional sectors.
Decomposition analysis reveals that wage benefits are concentrated among employees aged 45 and above, managers, and white-collar workers; other worker categories experience stagnant wages, and no group shows a negative wage effect.
Decomposition of wage effects by worker groups (age, occupation/type) using the integrated dataset and the DiD/other regression analyses.
Wage increases at small firms primarily explain the positive adoption effect, while wages at medium and large firms remain stagnant after adoption.
Heterogeneity analysis by firm size within the DiD framework showing differential post-adoption wage trajectories for small versus medium/large firms.
Non-routine employment and wages exhibit a crossing pattern: initially higher under fast adoption, then lower — so faster adoption can simultaneously raise long-run wages for survivors while permanently reducing participation.
Comparative dynamic trajectories in the model showing time paths for non-routine employment and wages under fast vs. slow adoption scenarios (analytical and/or simulated model paths).
The study establishes statistically significant relationships between organizational AI adoption and compensation dynamics.
Econometric estimates (difference-in-differences and propensity score matched comparisons) using the combined datasets listed in the paper and controlling for industry, firm size, geography, occupation characteristics, and macroeconomic variables.
Residual within-task group dynamics dominate the magnitude of the gender wage gap, though task-based employment and wage channels are important for timing and direction of changes in gender inequality in the formal sector.
Decomposition analysis partitioning the gender wage gap into within-task residuals and task-based employment and wage components, with residuals accounting for the largest share of the gap but task channels explaining temporal shifts.
The proportion of consumers who adopt AI-induced services influences the pricing of those services and through price adjustments will further impact wages across traditional and non-traditional services.
Theoretical development and analysis in the paper via a demand-switching model and a Finite Change General Equilibrium framework introducing AI as a technological shock modeled through price adjustments.
People compensate workers less for AI-assisted output even when output quality is held constant, because the work is perceived as requiring less effort and reflecting less of the worker's own agency.
The paper cites research on compensation for workers who use AI [11]. The underlying study's sample size and quantitative effect are not reported in the paper.
Clients pay freelancers less for validating and correcting AI-generated work than they previously paid for producing original work.
The paper presents this as an economic consequence of the producing-to-validating shift and supports it with evidence from machine-translation post-editing and research on the AI compensation penalty [11, 14]. No sample size or pooled effect estimate is reported.
Freelancers in occupations more exposed to generative AI experienced losses in both contracts and earnings, and experienced or top-rated freelancers did not fare better than other freelancers; the paper describes suggestive evidence that they were hit hardest.
The claim is based on cited research examining an online labor market [8]. The paper does not report the underlying study's sample size or quantitative effect estimates.
Most older adults in Korea's Senior Employment Program remain in simple-labor positions and receive monthly stipends of approximately KRW 270,000.
The paper cites Ministry of Health and Welfare (2025) program data.
Sheltered-workshop wages for persons with disabilities in Korea averaged approximately KRW 500,000 per month.
The paper cites Ministry of Employment and Labor (2024) and reports the wage figure as evidence of disadvantaged labor-market conditions.
Informal employment, weak promotion systems, low worker bargaining power, opaque recruitment, and non-transparent wage-setting can prevent productivity gains from translating into higher wages.
Institutional and labor-market analysis in the conceptual model; no randomized evaluation or large matched employer–employee dataset is reported.
The negative association between AI adoption and the high-skilled/low-skilled wage-cost ratio is concentrated in regions located in countries that are highly specialized in AI technologies.
Heterogeneity analysis comparing regional labor-market effects by the AI specialization of the countries in which regions are located.
The reduction in the high-skilled/low-skilled wage-cost ratio is broadly split between a decline in relative wages and a decline in the relative employment share of high-skilled workers.
The authors decompose the AI-associated change in the wage-cost ratio into wage and employment margins using their CES production framework and regional labor-market estimates.
AI adoption is associated with an approximately 7% reduction in the relative wage-cost ratio of high-skilled to low-skilled workers across European regions.
Empirical estimates based on European regional labor-market data and a CES production framework linking AI exposure to the high/low-skilled wage-cost ratio.
The paper states that the progressive integration of new technology expands the work executable by capital or automation, reducing labor's value addition and depressing wages and employment.
Theoretical and empirical claim attributed to Acemoglu and Restrepo (2024); the paper does not present the cited study's sample, estimates, or identification strategy.
In the simulated Track A panel, workers below the graduate level in AI-exposed occupations experienced a significant wage penalty after the onset of generative AI.
Difference-in-differences estimation using the simulated, PLFS-structured panel; the reported DiD coefficient was -0.126 with p < .001.
Politicians facing greater mutual oversight accumulate lower private returns from holding public office.
Analysis of sworn politician asset affidavits using the winner's premium, defined as the differential asset growth of election winners relative to runners-up; compares constituencies with different shares of split blocks.
Medium-skill routine workers show the strongest negative association between AI exposure and reporting higher household income.
Heterogeneity analyses across skill groups and occupation types using industry robot density as the measure of objective AI exposure.
A one-standard-deviation increase in industry robot density is associated with a 4.2 percentage-point lower probability that Chinese workers report higher household income than the previous year.
Ordered probit analysis of 6,247 employed respondents in the 2021 China General Social Survey, matched with industry-level robot penetration; effect reported as a predicted-probability contrast.
Within the study’s conceptual framework, Double Displacement threatens older highly skilled immigrants’ income security through reduced late-career earning capacity and a shortened recovery period before retirement.
Interpretive conceptual mapping of interview findings onto the WHO Active Ageing framework; the authors explicitly state that the study makes no causal claims.
Reduced visibility and networking opportunities for remote workers can harm career progression, including promotions, raises, and project assignments.
The paper identifies reduced workplace visibility and networking as mechanisms linking remote work to career outcomes; it does not report a quantified causal estimate.
Hui, Reshef, and Zhou are reported to find that the release of ChatGPT was associated with a 2% decline in the number of contracts and a 5.2% decline in monthly earnings in affected online-freelancing occupations, with top-rated providers hit at least as hard as other providers.
The paper summarizes a peer-reviewed difference-in-differences study of a large online freelancing platform.
Algorithmic management in gig and platform labor affects bargaining power, wage setting, monitoring intensity, and employment relationships.
Conceptual labor-economics analysis within the theoretical synthesis; no worker-level sample, wage estimates, or causal tests are reported.
Indonesia's legal framework falls short particularly in the area of income security for platform (ride-hailing) workers.
Analysis of Indonesian labour laws and regulations against the 'fair pay' Fairwork indicator; doctrinal/legal interpretation of statutory gaps related to income guarantees and social protections.
Although presented as optional, algorithmic recommendations create economic disincentives for non-compliance (i.e., ignoring recommendations reduces workers' economic returns).
Statistical analysis of worker logs and earnings dynamics using SLM to separate direct recommendation effects from spatial spillovers; reported as a causal impact of recommendation intensity on incentives.
Disclosure reduces aggregate creator surplus.
Welfare calculations in the formal model comparing creator surplus under disclosure versus non-disclosure regimes; shows aggregate creator surplus is lower under disclosure.
Vibe coding weakens user engagement through which many maintainers earn returns.
Mechanism described in the model (abstract): reduced direct engagement when users rely on AI assembly rather than interacting with maintainers.