Evidence (16496 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
9875 claims
Filter claims →
Productivity
8807 claims
Filter claims →
Governance
7870 claims
Filter claims →
Human-AI Collaboration
7560 claims
Filter claims →
Org Design
4892 claims
Filter claims →
Innovation
4781 claims
Filter claims →
Labor Markets
4004 claims
Filter claims →
Skills & Training
3308 claims
Filter claims →
Inequality
2332 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 | 870 | 233 | 116 | 1066 | 2363 |
| Governance & Regulation | 976 | 451 | 218 | 133 | 1809 |
| Organizational Efficiency | 949 | 224 | 144 | 88 | 1416 |
| Technology Adoption Rate | 764 | 287 | 141 | 122 | 1325 |
| Research Productivity | 501 | 152 | 74 | 362 | 1101 |
| Output Quality | 542 | 216 | 69 | 69 | 896 |
| Decision Quality | 387 | 198 | 94 | 54 | 740 |
| Firm Productivity | 513 | 67 | 101 | 27 | 714 |
| AI Safety & Ethics | 249 | 303 | 73 | 36 | 667 |
| Market Structure | 190 | 192 | 134 | 27 | 548 |
| Task Allocation | 243 | 77 | 91 | 36 | 452 |
| Innovation Output | 291 | 33 | 55 | 20 | 401 |
| Skill Acquisition | 206 | 72 | 65 | 21 | 364 |
| Employment Level | 133 | 63 | 115 | 22 | 335 |
| Fiscal & Macroeconomic | 153 | 79 | 52 | 32 | 323 |
| Task Completion Time | 206 | 37 | 12 | 15 | 272 |
| Firm Revenue | 179 | 52 | 29 | 5 | 266 |
| Consumer Welfare | 130 | 76 | 47 | 13 | 266 |
| Inequality Measures | 48 | 137 | 51 | 6 | 242 |
| Worker Satisfaction | 101 | 81 | 25 | 13 | 220 |
| Error Rate | 84 | 110 | 11 | 5 | 210 |
| Wages & Compensation | 98 | 47 | 30 | 10 | 185 |
| Regulatory Compliance | 88 | 73 | 17 | 7 | 185 |
| Automation Exposure | 66 | 64 | 33 | 16 | 182 |
| Team Performance | 105 | 29 | 30 | 11 | 176 |
| Training Effectiveness | 109 | 22 | 14 | 21 | 168 |
| Developer Productivity | 114 | 21 | 14 | 8 | 158 |
| Job Displacement | 12 | 90 | 24 | 1 | 127 |
| Hiring & Recruitment | 57 | 9 | 9 | 5 | 80 |
| Skill Obsolescence | 6 | 56 | 9 | 1 | 72 |
| Social Protection | 43 | 17 | 8 | 2 | 70 |
| Creative Output | 35 | 21 | 9 | 4 | 70 |
| Labor Share of Income | 18 | 21 | 17 | 1 | 57 |
| Worker Turnover | 15 | 16 | — | 4 | 35 |
| Industry | — | — | — | 1 | 1 |
Wan generates 66.5% of videos in the dataset.
Model-attribution analysis of the video subset of the collected SNEACI items; the paper reports Wan as the generator for 66.5% of videos.
Open-source models dominate production, with the Stable Diffusion family generating 42.7% of images.
Model-attribution analysis of the collected SNEACI items; the paper reports that 42.7% of images in the dataset were generated by the Stable Diffusion family.
Non-celebrity individuals now account for 55.8% of targets, compared to only 4.7% in prior studies.
Analysis of the identified dataset of 24,105 SNEACI items and comparison to figures reported in prior literature (prior studies reported 4.7% non-celebrity targets).
We identify 24,105 SNEACI items.
Large-scale empirical data collection and content identification in the anonymous content community; the paper reports locating and labeling 24,105 synthetic non-consensual sexually explicit imagery (SNEACI) items.
Traditional dynamic pricing models in large-scale e-commerce suffer from limited interpretability, poor utilization of unstructured information, and misalignment with long-term business objectives such as cumulative Gross Merchandise Value (GMV), Return on Investment (ROI) and milestone achievement.
Author statement in paper abstract/introduction asserting limitations of existing dynamic pricing models (literature/motivation claim); no empirical test or sample size reported in the provided text.
The review identifies significant compliance challenges related to emerging regulations, including New York City Local Law 144, Illinois HB 3773, and the European Union AI Act.
Legal and policy scholarship included in the systematic review of 34 studies, which discuss regulatory requirements and compliance issues associated with AI-based recruitment.
AI-based recruitment systems frequently inherit demographic and historical biases embedded within training datasets, potentially leading to discriminatory hiring outcomes when adequate oversight mechanisms are absent.
Synthesis of findings across the systematic review of 34 studies reporting evidence of demographic/historical biases in training data and downstream discriminatory effects in hiring models.
There is a global disparity in data centre infrastructure (concentrations favouring some regions over others).
Analysis drawing on external data sources cited in the paper illustrating geographic distribution of data centre infrastructure.
Data workers in Kenya report direct employment by big tech corporations and exposure to graphic content.
Qualitative interviews / responses from data workers in Kenya collected and reported in the paper.
Hyper-datafication systematically redistributes labour risks and representational harms toward the Global South.
Qualitative responses from data workers in Kenya describing labour conditions and exposure; analysis of language data representation; external data on global data centre infrastructure and geography.
Hyper-datafication drives substantial and growing environmental costs.
Quantitative analysis of dataset growth and estimated storage-related energy consumption and carbon footprint across the analysed Hugging Face datasets (≈550k); modelled storage and emissions impacts.
Türkiye is one of the most fragile regimes due to its weak regulatory capacity, high algorithmic discipline, and lack of transparency.
Regime assessment in the comparative analysis component of the paper that evaluates Türkiye's regulatory capacity and algorithmic governance characteristics.
Regime positioning reveals that despite the EU's partial regulatory capacity, it cannot fully close the collective rights gap.
Comparative normative analysis of EU regulatory frameworks relative to collective algorithmic rights dimensions (paper's regime positioning assessment).
Individual-centered regulatory frameworks (GDPR, AI Act, CCPA, LGPD, etc.) are limited in their understanding of the collective operating logic of algorithmic governance.
Normative comparative analysis of existing regulations across the EU, US, Latin America, Asia, and Türkiye as reported in the paper (conceptual/legal analysis rather than empirical measurement).
Algorithmic governance under a data-driven, predictive, and dynamic authority architecture is creating structural transformations that exceed the institutional capacity of the existing individual rights paradigm.
Conceptual argument presented in the paper; theoretical analysis of algorithmic governance and its impacts on institutional frameworks (no empirical sample reported).
Profitability has a negative effect on earnings management (more profitable firms engage in less earnings manipulation).
Regression results showing a statistically significant negative coefficient for profitability in models predicting earnings management.
AI adoption is strongly negatively correlated with earnings management (discretionary accruals): greater AI use in auditing is associated with lower discretionary accruals.
Empirical results from panel regressions on sample of 680 firm–period observations showing a strong negative correlation between AI usage measures and discretionary accrual measures.
Re-asking the same GPQA-Diamond questions in free-response rather than multiple-choice form reopens the all-wrong tail: beta is 0.127; a five-judge panel had kappa 0.73 to 0.92, locating co-failure in answer format rather than subject.
Empirical comparison of GPQA-Diamond questions in multiple-choice vs free-response format, with human adjudication by a five-judge panel and reported inter-rater agreement (kappa range).
On execution-graded code, the observed all-wrong rate beta is 0.079.
Empirical measurement on execution-graded code tasks reported in the paper (dataset/method described in the study).
Across 67 models from 21 providers on open-ended mathematics, observed beta is 0.052 versus 0.023 under the full 67-model Gaussian copula (tetrachoric-calibrated single-factor model underprices the all-wrong tail), about 2.5 times underpricing with 90% CI 1.7 to 3.4 and k = 17.
Empirical evaluation across 67 models from 21 providers on open-ended mathematics; comparison between observed all-wrong rate and rate predicted by a full 67-model Gaussian copula / tetrachoric-calibrated single-factor model; confidence interval reported and k specified (k = 17 as reported).
For any policy whose output is one member model answer, accuracy cannot exceed one minus beta, where beta is the rate at which every model is wrong on the same query.
Theoretical derivation/proof given in the paper (a formal upper bound linking ensemble accuracy to the all-wrong rate beta).
This analysis (identifying robust trait combinations across the full tested space) was infeasible with the mechanistic model alone.
Comparative statement: inability to perform such exhaustive search with APSIM due to computational expense, contrasted with capability of the emulator.
AI has a significant negative influence on value chain upgrading in labor-intensive equipment manufacturing industries.
Industry-type heterogeneity analysis within the same 30-province panel (2010–2022) showing a statistically significant negative coefficient for labor-intensive subsectors.
This convergence has the potential to lower wages on entry-level thinking jobs.
Theoretical/empirical implication drawn from observed reduction in productivity differences; presented as a potential consequence rather than an established empirical result in the abstract.
Early evidence indicates AI is reducing the productivity difference between beginner and expert employees.
Reported 'early evidence' from the paper's empirical analysis (difference-in-differences on freelance platforms) indicating convergence in productivity between novices and experts; no numeric effect estimates given in the abstract.
There is a Verifier-Goodharting Floor on flywheel ceilings under imperfect rewards (a formal result showing a lower bound / floor imposed by reward imperfection and verifier limitations).
Formal theoretical result in the paper (derivation showing that imperfect/verifier-mediated rewards create a floor limiting flywheel / self-improvement ceilings).
There exists a Determinism-Efficiency Bound on chain-task success (one of three formal results pinning down the regime).
Formal theoretical result presented in the paper (a derived bound relating environment determinism to chain-task efficiency/success).
With per-step determinism δ < 1, k-step chain success degrades as δ^k (long-chain agent execution fails exponentially in environments designed for human tolerance).
Formal theoretical statement in the paper: a mathematical characterization of per-step determinism and k-step chain success (Derivation / formal result presented by the authors).
Infrastructural deficiencies, unstable electricity supply, limited technical expertise, and high implementation costs remain major barriers to AI adoption.
Survey responses from 522 participants reporting barriers to AI adoption; descriptive analysis identifying these factors as major barriers.
The study identifies feedback loops and conditions for systemic instability, including potential tipping points in environmental and financial regimes.
Analytical/theoretical identification of feedback structures and instability conditions within the model; no empirical testing or sample reported.
Artificial intelligence simultaneously increases system coupling and transition sensitivity.
Theoretical claim derived from synergetic theory and the paper's model; no empirical evidence or sample size provided.
A strategic AI sender may withhold evidence or garble information in order to steer the human's decision.
Theoretical reasoning and examples within the Bayesian persuasion framework showing that sender-optimal signaling need not fully reveal the state and can be manipulative; supported by model analysis rather than empirical data.
Within the scope studied, these results sit in tension with a strong winner-takes-all narrative around AI recommendation.
Synthesis of empirical findings (moderate Gini, low incidence of vacuums, cross-model disagreement, displacement ratios) contrasted with the expectation of winner-takes-all concentration.
Competitive vacuums were rare, appearing in 8.0% of queries; models named at least one sampled brand in most cases.
Empirical count/proportion of queries classified as competitive vacuums using the Competitive Vacuum Index across the study's queries.
Recommendation concentration was moderate in the sample: the mean Gini coefficient was 0.28 (95% CI [0.16, 0.41]), below the 0.60 power-law threshold the authors set.
Empirical calculation of Gini coefficients across categories using the study's sample (implied across the 250 brand-free category queries and associated responses).
These results demonstrate how people's decision-making processes can be insufficient for overseeing AI in high-stakes domains.
Synthesis/interpretation of experimental findings (longer viewing when no AI, small increases in selection probability with more time for non-recommended candidates, IAT effects) to argue that human decision processes may not adequately supervise biased AI in high-stakes settings. This is an interpretive/concluding claim based on the experiment; not a direct empirical measure. Sample size not stated in the excerpt.
Prior research has largely focused on established firms, with limited attention to startups.
Literature review / gap statement in the paper asserting an imbalance in existing research coverage (no quantitative meta-analysis reported).
In manual jobs, AI compresses the returns to undereducation as tasks become more skill-intensive.
Occupation-specific heterogeneity analysis using CLDS and city AI diffusion showing reductions in the undereducation wage premium within manual-occupation subsamples under higher AI diffusion.
AI diffusion slightly lowers the wage premium for undereducated workers.
Interaction effects from fixed-effects models using CLDS and city AI diffusion indicators showing a small reduction in undereducation-related wage premium with higher AI diffusion.
Overeducation leads to a significant wage penalty.
Microdata from the China Labor-force Dynamics Survey (CLDS) 2014–2018; cohort-based measure of educational mismatch; estimated using extensive fixed-effects models comparing wages by educational mismatch status.
Pooled across five AI coding agents, pull requests (PRs) with a human Co-Authored-By trailer merge less often than purely-autonomous ones (53.8% vs. 79.8%).
Aggregate analysis of PR merge rates across five AI coding agents in the AIDev dataset; pooled sample of PRs (33,596 PRs referenced elsewhere in the paragraph).
The Board Interlock Network (BN) acts as a key informal governance mechanism that, together with media attention, produces a synergistic governance effect inhibiting speculative AI-related disclosure.
Theoretical framing and empirical results (causal DML estimates, mechanism tests) presented in the paper highlighting the interaction between internal governance (interlocking directorates) and external information intermediaries (media).
The inhibitory effect of board interlocks on AI Wash is more pronounced in firms facing stronger external monitoring pressures, such as intense market competition.
Heterogeneity analysis reported in the paper comparing subsamples based on market competition intensity, indicating larger BN effects under greater competition.
The inhibitory effect of board interlocks on AI Wash is more pronounced in firms with more complex governance needs, such as state-owned enterprises (SOEs).
Heterogeneity analysis reported in the paper comparing subsamples (e.g., SOEs vs. non-SOEs) showing stronger BN effects in SOEs.
The inhibitory effect of board interlocks on AI Wash is more pronounced in firms with greater information asymmetry, for example, high-technology firms.
Heterogeneity analysis reported in the paper comparing subsamples (e.g., high-technology vs. others) showing stronger BN effects in high-tech firms.
The finding that the BN inhibits AI Wash remains robust after a series of robustness and endogeneity tests.
Reported robustness checks and endogeneity tests applied to the DML estimation on the Chinese A-share sample (2007–2022).
The Board Interlock Network (BN) significantly inhibits corporate AI Wash.
Causal inference using a Double Machine Learning (DML) model on a sample of Chinese A-share listed companies from 2007 to 2022; robustness and endogeneity tests reported.
The paper identifies two distinct gaps that have widened as GPTs exposure scores traveled from their time and place of production: (1) a structural gap between what static exposure scores measure and what policy questions require, and (2) a coordination gap between researchers and policymakers.
Explicit framing and thesis presented in the paper summarizing the central arguments.
Policy-relevant work that asks who is harmed or benefits, how, and when continues to reference static GPTs exposure scores without engaging with methodological updates needed to answer these questions more reliably.
Critical literature review and observed citation practices reported by the authors; claim based on review of how policy analyses cite/ use the scores.
These temporal, geographic, and ontological limitations compound when exposure scores are used in policy-facing analyses.
Conceptual argument and case-study approach in the paper showing how limitations interact and worsen policy analysis outcomes.