Evidence (426 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
21267 claims
Filter claims →
Productivity
17978 claims
Filter claims →
Governance
17038 claims
Filter claims →
Human-AI Collaboration
16914 claims
Filter claims →
Org Design
11104 claims
Filter claims →
Innovation
11087 claims
Filter claims →
Labor Markets
6711 claims
Filter claims →
Skills & Training
5616 claims
Filter claims →
Inequality
4343 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 | 1880 | 496 | 296 | 1854 | 4721 |
| Organizational Efficiency | 2906 | 665 | 438 | 180 | 4210 |
| Governance & Regulation | 2162 | 929 | 480 | 247 | 3866 |
| Technology Adoption Rate | 1533 | 545 | 278 | 210 | 2593 |
| Decision Quality | 1391 | 534 | 321 | 173 | 2429 |
| Output Quality | 1298 | 472 | 231 | 145 | 2153 |
| AI Safety & Ethics | 682 | 821 | 230 | 90 | 1837 |
| Research Productivity | 855 | 253 | 121 | 425 | 1675 |
| Firm Productivity | 1105 | 171 | 175 | 73 | 1531 |
| Task Allocation | 735 | 229 | 361 | 99 | 1433 |
| Market Structure | 457 | 461 | 251 | 47 | 1222 |
| Innovation Output | 673 | 94 | 108 | 36 | 913 |
| Task Completion Time | 499 | 118 | 43 | 38 | 702 |
| Firm Revenue | 458 | 130 | 61 | 26 | 677 |
| Skill Acquisition | 381 | 122 | 113 | 34 | 650 |
| Consumer Welfare | 316 | 176 | 115 | 39 | 648 |
| Employment Level | 223 | 143 | 177 | 53 | 600 |
| Error Rate | 246 | 282 | 44 | 19 | 594 |
| Fiscal & Macroeconomic | 283 | 142 | 78 | 52 | 562 |
| Inequality Measures | 103 | 329 | 106 | 13 | 552 |
| Worker Satisfaction | 225 | 185 | 63 | 30 | 503 |
| Automation Exposure | 158 | 155 | 72 | 37 | 426 |
| Regulatory Compliance | 186 | 126 | 35 | 14 | 362 |
| Team Performance | 193 | 56 | 51 | 24 | 326 |
| Developer Productivity | 224 | 58 | 27 | 13 | 323 |
| Wages & Compensation | 148 | 108 | 50 | 17 | 323 |
| Training Effectiveness | 218 | 44 | 21 | 27 | 313 |
| Job Displacement | 23 | 159 | 53 | 5 | 240 |
| Hiring & Recruitment | 109 | 61 | 32 | 11 | 215 |
| Skill Obsolescence | 16 | 107 | 26 | 6 | 155 |
| Creative Output | 71 | 44 | 28 | 6 | 150 |
| Social Protection | 58 | 31 | 12 | 3 | 104 |
| Labor Share of Income | 29 | 43 | 25 | 2 | 99 |
| Worker Turnover | 45 | 29 | 6 | 4 | 84 |
| Industry | — | — | — | 1 | 1 |
Adoption intensity has an inverted-U relationship with incidents in the AI Incident Database, with an implied peak near 73% adoption.
Real-data panel analysis of 1,383 AI incidents from the AI Incident Database covering 2019–2026.
Augmentation AI exposure is primarily concentrated in STEM occupations, whereas automation AI exposure is relatively more prevalent in high-skilled occupations and in sales, office, and administrative-support occupations.
Descriptive analysis of occupational exposure indices constructed from AI-related Stack Overflow questions mapped to occupational abilities and outputs.
AI increasingly affects cognitive and analytical functions traditionally associated with highly educated workers, giving it the potential to influence virtually every sector of the economy.
Conceptual comparison of AI with earlier waves of automation in the introduction; the essay provides sectoral examples but no empirical sector-by-sector analysis.
AI and task automation increase substitutability most strongly for work that is decomposable and codifiable, whereas AI augmentation can counterbalance this effect by increasing workers' verifiable productivity.
Theoretical synthesis of AI-enabled knowledge-work studies and global labour arbitrage literature; no task-level experiment or observational estimate is reported.
AI increasingly affects cognitive and analytical functions traditionally associated with highly educated workers, giving it the potential to influence virtually every sector of the economy.
Conceptual comparison of AI with earlier waves of automation in the introduction; the essay provides sectoral examples but no empirical sector-by-sector analysis.
The study finds that generative AI may automate structural creative tasks while leaving relational and affective tasks comparatively resistant to automation, producing hybridized job profiles rather than wholesale displacement.
Interpretation of the reported generational workflows and participants’ distinction between structural AI-assisted tasks and human relational work; this is an implication of the qualitative findings, not a measured labor-market effect.
Higher AI intensity increases both firms’ regulatory exposure and the opportunities that skilled legal capability can unlock.
Conceptual reasoning concerning AI-related privacy, safety, liability, data protection, algorithmic bias, and platform-regulation issues.
The strength of psychological barriers to enterprise AI adoption differs by enterprise size, industry type, and employees' prior AI experience.
Multi-group comparison analyses examined heterogeneity across enterprise and employee subgroups in a three-enterprise survey.
The six AI systems differed more across cognitive capability dimensions than across model families.
Comparative profiling of six AI systems using the paper's cognitive-capability battery.
Industrial AI and digital-twin systems in the Vietnamese electric-vehicle welding plant do not fully automate or simply replace human labor; their operation depends on workers’ routine desynchronization of machine-directed processes.
Qualitative ethnographic fieldwork combining shop-floor observation and semi-structured interviews at a single high-tech electric-vehicle welding factory in Vietnam.
Occupations requiring complex interpersonal judgment, caregiving, and pedagogical tasks generally have lower AI exposure, although individual tasks within these jobs may still be automated or augmented.
Comparative occupational exposure classification across healthcare and education, based on task content and AI adoption indicators.
The framework is focused on the customs-observable subset of TBML, while many TBML techniques remain outside its scope or require cooperation from financial gateways, registries, or international partners.
Scope limitations and detectability mapping; no empirical estimate of the proportion of TBML covered is reported.
The paper maps ten TBML techniques to their observable data sources, detectable architectural layers, and blind spots, making explicit which techniques are amenable to customs-observable screening.
Typology-detectability scope mapping across customs, registry, and financial-gateway observables.
The IMF estimates that almost 40% of global employment is exposed to AI, with exposure of 60% in advanced economies, 40% in emerging economies, and 26% in low-income countries.
The paper cites IMF cross-country estimates of employment exposure to AI.
The fraction of intervened turns without a policy-supported counterfactual was substantially higher for Llama-3.1-8B than for Qwen2.5-7B: 26.8% versus 13.1%, a factor of 2.05.
Replay under the same environment, four alternatives, and 15-rollout budget; Qwen had 2,034 intervened turns and Llama had 1,082 corrected-instrument turns.
Topology can shift aggregate reliance when network influence transmits correlated beliefs rather than exchangeable experience signals.
Boundary experiment comparing experience-based learning with opinion dynamics and varying whether high-degree hubs initially had high or low trust.
In a 32-step simulated corporate-network task, Kimi K3 reached step 17 on average and completed the scenario once in ten attempts, compared with step 28.5 and six or seven successful attempts in ten for the strongest closed models.
Joint AISI/CAISI evaluation of Kimi K3 against leading closed models in a simulated corporate network.
On the ExploitBench benchmark, the open-weight Kimi K3 model scored 32%, compared with 24% for the previous most capable open-weight model, but it remained substantially behind closed frontier models.
Joint UK AI Security Institute and US Center for AI Standards and Innovation assessment on 41 recent V8-engine vulnerabilities.
Regional exposure and local AI intensity are important sources of heterogeneity in AI's labor-market effects.
Regional comparisons and heterogeneity analysis based on differences in AI exposure and national AI specialization.
The study’s 63-competency taxonomy consists of 28 competencies classified as high-AI-potential displacement competencies and 35 classified as moderate-AI-potential augmentation competencies.
An ensemble classification protocol used Gemini 1.5 Pro, ChatGPT-4o, Claude 3.5 Sonnet, Microsoft Copilot, and Perplexity Pro, followed by consensus reconciliation; 57 low-AI-potential competencies were excluded.
Across 846 U.S. occupations, the AI Dual-Track model identifies simultaneous displacement and augmentation pressures associated with generative AI.
Occupational-level analysis merging O*NET competency data with BLS wages and employment data for a final sample of 846 occupations; the model assigns competencies to substitution and facilitation tracks.
Structural inequalities, governance quality, and development conditions shape both mobility choices and households' capacities to adapt in place.
Comparative and case-focused analysis of environmentally vulnerable regions combined with literature and governance analysis.
Generative AI lowers the marginal cost of content production, affecting creative labor markets, journalism business models, and the supply of both high-quality and deceptive content.
Qualitative analysis of production structures and economic implications; no labor-market estimates are reported.
Germany, France, Estonia, and Poland occupy intermediate vulnerability positions, but the dimensions driving vulnerability differ across countries.
The comparative results describe labour-market exposure as important in Germany and France, migration pressures in Estonia and Poland, and limited technological readiness particularly in Poland.
The comparative validation sample consists of six countries selected using cluster analysis: Germany, France, Finland, Estonia, Poland, and Ukraine.
Country selection used cluster analysis to identify a representative six-country sample spanning different vulnerability patterns.
AI exposure is a technological possibility and should not be interpreted by itself as a prediction of adoption, substitution, job loss, or welfare effects.
Review of task-, occupation-, skill-, firm-, and time-varying measures of AI exposure, followed by a conceptual distinction between exposure and realized economic outcomes.
Yapay zekâya maruziyet büro, yönetim ve diğer bilişsel yoğunluk taşıyan beyaz yaka mesleklerde daha yüksek; fiziksel emek ağırlıklı ve düşük vasıf gerektiren mesleklerde ise görece daha düşüktür.
Georgieff ve Hyee (2022) tarafından bildirilen meslekler arası yapay zekâ maruziyeti karşılaştırmaları.
The paper presents Hugging Face game-asset model release velocity as a candidate leading indicator of production-cost decline, but does not establish a lagged causal or predictive relationship.
Four annual observations from an API-search-limited sample; the paper states that no lag structure can be estimated and calls for quarterly-resolution validation.
This study develops a disruption index that integrates task exposure, adoption rates, time savings, and skill complementarity.
Methodological contribution described in the paper (index construction combining specified components).
AI integration does not simply automate entrepreneurial work.
Stated as a central argumentative claim in the thesis excerpt; no empirical methods, data, or sample size provided in the excerpt.
By identifying differences across male and female-dominated occupations, this work supports recommendations to mitigate the risk of reinforcing gender inequalities in the labour market.
Paper claims its occupational exposure analysis distinguishes male- and female-dominated occupations and uses those findings to support policy/recommendation proposals; the excerpt does not provide the analytic method, magnitude of differences, or sample details.
We present evidence from a Responsible Artificial Intelligence (RAI) UK project examining the exposure of different occupations to AI-driven innovation.
The paper states it provides empirical evidence from a RAI UK project analyzing occupation-level exposure to AI-driven innovation; methods and sample size are not included in the provided excerpt.
The transformation of Turkiye's digital newsrooms is best characterized as AI-assisted rather than AI-led.
Author interpretation/argument synthesizing survey findings (role composition, task uses, attitudes) to characterize the nature of AI integration.
Exposure to GenAI is concentrated in high-income economies, where 34% of employment is in exposed occupations, compared with 11% in low-income countries.
Cross-country comparison of employment shares in exposed occupations using ILO data and Working Papers 96 and 140.
Aggregating task-level AI exposure scores to the job-level reveals heterogeneity in AI exposure (mean and variance) even for seemingly identical jobs.
Analysis approach: aggregation of task-level exposure scores to compute job-level means and variances; paper asserts this reveals heterogeneity among ostensibly identical roles.
We find marked heterogeneity in model predictions.
Empirical comparison of multiple projection models (comparison of outputs across the set of models described in the paper).
Youthful populations, large shares of informal and agricultural employment, and concentrated digital and outsourcing hubs create a patchwork of exposure: pockets of intense AI adoption sit beside vast swathes of low-digitisation employment.
Conceptual synthesis and descriptive evidence cited from major international organizations and industry studies (World Bank, ILO, etc.) in the paper; no numerical sample provided.
Both full-panel estimates (hiring and productivity) are imprecise, pointing to augmentation rather than displacement.
Author discussion noting imprecision of full-panel coefficients and interpreting the pattern as consistent with augmentation (productivity up, hiring down slightly but imprecise).
AI exposure varies sharply across Indian firms.
Author statement based on firm-level exposure measures computed in the paper (weighted averages of occupational LLM exposure scores across business lines).
The dominant paradigm has shifted from 'substitution' (machines replacing workers) to 'augmentation' (AI augmenting human work).
Interpretive conclusion in the paper drawn from secondary literature (WEF, ILO, McKinsey, PwC) and observed policy/industry trends.
AI exposure is more positive for occupations performing nonroutine interactive work and more negative for occupations concentrated in analytical, scientific, and operations-control skills.
Occupation-level analysis mapping skill content (interaction-and-communication vs. analytical/scientific/operations-control) to market-implied AI premium; comparison across occupational skill categories.
There is a growing reliance on agentic AI systems within the platform context.
Qualitative evidence from the 20 interviews and the 24-participant workshop reporting increased dependence on AI agents for tasks and decision support.
There is increasing automation of operational tasks in the development domain.
Participant reports and workshop discussions from 20 interviews and a 24-person workshop indicating automation of operational activities; qualitative thematic evidence.
The intended contribution is an Information Systems framework explaining when AI supports human augmentation and when it produces functional substitution.
Stated intended theoretical contribution in the abstract (proposed framework). This is an intended outcome rather than an empirically demonstrated result in the provided text.
The present wave of automation targets non-routine cognitive activity such as coding, technical writing, and graphic design, unlike past automation which mainly involved routine manual activity.
Framing/background statement in the paper contrasting historical automation (routine manual tasks) with current AI-driven automation of non-routine cognitive tasks; no sample size or quantitative test reported in the abstract.
AI's rapid evolution has profound effects on the labor market, influencing the levels, skills needed for jobs, and overall jobs content.
Statement from the paper's synthesis/introduction summarizing reviewed empirical studies (systematic literature review covering studies from 2017–2025). Number of underlying studies not reported in the excerpt.
Although the geometry (bipolar structure) is stable, its content is not: across a decade the polarity has inverted relative to Frey and Osborne (2013).
Comparison of macro-level placements between the paper's LLM-era OAI and the Frey-Osborne (2013) rankings; authors report inversion and supporting correlation statistics.
Tool-Mediated Physical (M2) and Planning & Design (M7) are separated by Cohen's d = 2.41 (H = 172.88, p = 6.21e-34).
Statistical comparison reported in the paper (Cohen's d, H-statistic, p-value) between the two macro clusters' OAI distributions.
Projecting the DWA-level Occupational Automation Index (OAI) onto a 7-macro semantic typology produces a bipolar structure (two poles separated by a low-contrast middle band).
Authors' projection of previously computed DWA-level OAI onto a 7-cluster semantic typology and subsequent analysis of cluster structure.
Artificial Intelligence (AI) has changed how people work across various fields and businesses, especially in the Indian Information Technology (IT) industry.
Authors' qualitative synthesis of peer-reviewed literature and thematic evaluation of secondary data (literature review). No sample size reported.