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 |
The findings show clear and systematic differences in how regions govern AI.
Comparative analysis of coded policy documents (n=24) producing indices that the authors interpret as showing systematic cross-regional differences in governance approaches.
The documents are systematically coded across four institutional dimensions and converted into simple indices to compare governance approaches across the regions.
Author-reported method: systematic coding of documents on four institutional dimensions and construction of indices for cross-regional comparison (based on the 24 documents).
This study uses a comparative qualitative policy analysis based on 24 key AI policy documents published between 2018 and 2025 across the European Union, United States, China, and Indo-Pacific economies.
Author-stated research design and sample: systematic review/comparative qualitative policy analysis of 24 AI policy documents spanning 2018–2025 covering EU, US, China and Indo-Pacific economies.
Actual sharing often contradicted willingness to share (the privacy paradox), with consistently high sharing rates across all conditions.
Observed discrepancy reported in the experimental results (N=240): despite variation in willingness-to-share, behavioral sharing rates remained high and similar across human, white-box AI, and black-box AI conditions.
Energy policy uncertainty has a nonlinear effect on AI investment: moderate uncertainty fosters innovation, whereas high volatility hinders long-term investment.
Empirical analysis using nonlinear methods (WQR and WQC) on US quarterly data 2013Q1–2024Q4 (48 quarters), assessing distributional asymmetries across quantiles and time–frequency bands.
Machine-readable metrics and open scholarly infrastructure are reshaping scholarly profiles and incentives.
Conceptual and historical discussion referring to platforms and metrics (e.g., arXiv, Google Scholar, ORCID) as mechanisms changing incentives; no new empirical estimates provided.
That interconnected ecosystem is fundamentally restructuring who can do science (access), how fast discoveries propagate, and what counts as a valid scientific contribution.
Argumentative claim linking infrastructural and tool changes to changes in access, dissemination speed, and norms of contribution. The paper presents examples and narrative but no systematic empirical evaluation or sample.
The most consequential development is not any single tool but the emergence of an interconnected ecosystem—AI agents, preprint platforms, open source codebases, and citation infrastructure—that forms a feedback loop.
Synthesis/argument based on multiple examples (LLM agents, preprint servers like arXiv, open-source code repositories, citation indices). No quantitative measurement or causal identification reported.
The central tension in AI for science is between automation (building systems that replace human researchers) and augmentation (tools that amplify human creativity and judgement).
Analytical claim based on the paper's review of historical examples and conceptual discussion; no primary data or experimental design reported.
Science has repeatedly delegated its bottlenecks to machines—first inference, then search, then measurement, then the full workflow—and each delegation solves one problem while exposing a harder one underneath.
Interpretive historical argument drawing on examples across AI-for-science milestones (e.g., DENDRAL, search and inference systems, measurement automation, and contemporary end-to-end workflows). No quantitative sample or experimental method reported.
Testing revealed AI excels at computational tasks but consistently misses nuanced factors like new construction rent premiums and infrastructure proximity impacts, validating the framework's hybrid structure as essential for professional-grade underwriting.
Findings from the controlled ChatGPT-4 test on the single 150-unit scenario: qualitative and comparative observations showing AI handled computations well but failed to capture specific local-market nuances, leading authors to endorse a hybrid human-AI framework.
Phase Two requires human-led professional validation to correct AI limitations, apply local market knowledge, and integrate risk factors.
Framework description supported by observations from the controlled test where human review was used to correct AI outputs and apply local knowledge (e.g., adjusting for nuanced market factors).
The growth effects of AI are conditional on institutional quality and organizational adaptability.
Theoretical/analytical claim in the paper's framework and supported by the stylized-facts analysis indicating heterogeneity in productivity and growth outcomes by institutional and digital capacity indicators.
AI agents implicate many areas of law, ranging from agency law and contracts to tort liability and labor law.
Legal/policy analysis in the paper enumerating legal domains implicated by AI agents (qualitative analysis; no sample size).
Firms of different ownership structures and industries exhibit different responses to the income distribution changes brought by AI (heterogeneous effects).
Paper reports performing grouped regressions by ownership type and industry to identify heterogeneous responses.
Financing constraints are a key factor that hinder firms' choice of technology level, which alters the corresponding income distribution effect of AI.
Paper posits financing constraint as a moderator and states it is considered in empirical analysis (interaction/moderation tests).
The development of AI may trigger new changes in the interest pattern between corporate profits and labor compensation.
Framed as the central research question/hypothesis; paper conducts empirical tests on firm panel data to evaluate this.
Artificial intelligence is profoundly reshaping the organizational form, operating model and operating mechanism of enterprises, and bringing unprecedented impact to the income distribution structure within enterprises.
Statement asserted in the paper's introduction/abstract; motivates empirical analysis using panel data of Shanghai and Shenzhen A-share non-financial listed firms (2010–2022).
These findings contribute to the literature by providing empirical insights from a developing economy, where unique socioeconomic and institutional factors shape the impact of AI.
Scope/claim of contribution based on the study's context (Cambodia) and its dataset (n = 351).
This study employed PLS‐SEM analysis on data from 351 respondents, revealing significant workforce reshaping.
PLS-SEM analysis conducted on survey data (n = 351) as reported in the paper.
The rapid adoption of artificial intelligence (AI) is fundamentally transforming labor markets worldwide, presenting both opportunities and challenges.
Statement made in the paper as background/justification; not based on the study's empirical data.
Traffic performance is sensitive to the distribution of safe time gaps and the proportion of RL vehicles.
Simulation results comparing Fundamental Diagrams across scenarios with different distributions of safe time gaps and shares of RL-controlled vehicles. Number of simulation runs or replicates not stated in the claim text.
Chat intent varies systematically with both the timing of chat relative to search and the category of products later purchased within the same journey.
Cross-tabulation/regression-style descriptive analysis relating classified chat intents to timing (relative to search) and subsequent purchased product categories in journey-level logs.
AI assistance in safety engineering is fundamentally a collaboration design problem rather than merely a software procurement decision: the same tool can either degrade or improve analysis quality depending entirely on how it is used.
Synthesis of the formal framework and analytic results in the paper (theoretical argument; no empirical sample reported).
The paper concludes by discussing open challenges in evaluating harmful manipulation by AI models.
Paper includes a discussion/conclusion section enumerating open challenges; stated in abstract.
We identify significant differences across our tested geographies, suggesting that AI manipulation results from one geographic region may not generalise to others.
Empirical comparison across three locales (US, UK, India) showing statistically significant differences in manipulation outcomes by geography.
Context matters: AI manipulation differs between domains, suggesting that it needs to be evaluated in the high-stakes context(s) in which an AI system is likely to be used.
Comparative analysis across three domains (public policy, finance, health) showing differences in manipulative behaviour and/or impact by domain in the empirical study.
AUROC_2 and M-ratio produce fully inverted model rankings, demonstrating these metrics answer fundamentally different evaluation questions.
Metric comparison across models showing that AUROC_2-based ranking and M-ratio-based ranking are fully inverted in the reported results on the evaluated dataset.
Temperature manipulation shifts Type-2 criterion while meta-d' remains stable for two of four models, dissociating confidence policy from metacognitive capacity.
Experimental manipulation (temperature changes) applied to models; reported result that Type-2 criterion shifted with temperature while meta-d' was stable for two models (out of four) in the 224,000-trial dataset.
Metacognitive efficiency is domain-specific, with different models showing different weakest domains, invisible to aggregate metrics.
Domain-level analyses reported in the paper showing per-domain M-ratio results and identification of different weakest domains per model, contrasted with aggregate metric behavior.
Metacognitive efficiency varies substantially across models even when Type-1 sensitivity is similar — Mistral achieves the highest d' but the lowest M-ratio.
Empirical comparison of Type-1 sensitivity (d') and metacognitive efficiency (M-ratio) across the four evaluated LLMs on the 224,000 QA trials; explicit statement that Mistral had highest d' but lowest M-ratio.
The paper's findings deepen the understanding of algorithmic aversion in the context of generative AI and offer practical guidance for creators and platforms navigating transparency versus engagement trade-offs.
Authors' interpretation and conclusions summarized in the abstract, based on the two experiments (study 1: n = 325; study 2: n = 371).
The paper's primary contribution is to combine established ingredients—attention scarcity, free-entry dilution, superstar effects, and preferential attachment—into a unified framework directed at claims about AI-enabled entrepreneurship.
Stated contribution and methodological description in the paper (synthesis and applied formalisation); this is a descriptive/methodological claim rather than an empirical result.
Modern pretrained time-series foundation models can forecast without task-specific training, but they do not fully incorporate economic behavior.
Statement in paper's introduction/abstract summarizing prior capabilities and limitations of pretrained time-series foundation models (no experimental sample or numeric evidence provided in the excerpt).
The governance risk-mitigation effects of AI operate through increasing financial risk exposure.
Authors' mechanism tests indicate a relationship between AI adoption and changes in financial risk exposure measures, which they interpret as a channel affecting executive behavior.
Organizational culture and technological readiness moderate the effectiveness of generative AI integration in decision-making processes.
The paper reports moderation effects tested in the SEM framework using survey data from senior managers, decision-makers, and AI adoption specialists (SmartPLS). No numeric moderator effect sizes or sample size provided in the excerpt.
The paper draws comparisons between inference tokens and established commodities such as electricity, carbon emission allowances, and bandwidth to motivate financialization.
Theoretical comparison and historical analysis (drawing on the historical experience of electricity futures markets and commodity financialization theory) as presented in the paper.
The effects of financial digital intelligence on the innovative development of strategic emerging industries vary across regions and sectors: there are differences across central, eastern, and western regions and across capital‑intensive and technology‑intensive sectors, while no significant impact is noted in other regions and industries.
Heterogeneity analysis reported on the panel dataset (5,731 observations, 2015–2022) examining regional and industry subsamples (details of subgroup sizes and statistical tests not provided in excerpt).
Initiatives such as Cassava AI's network of AI factories signal growing interest in adopting AI in Africa, but these projects remain very targeted and continental adoption still requires better coordination between African stakeholders.
Cited example (Cassava AI) in the paper to illustrate nascent initiatives; combined with the authors' qualitative assessment of scope and geographic targeting of such projects.
Small language models offer privacy-preserving alternatives to frontier models, but their specialization is hindered by fragmented development pipelines that separate tool integration, data generation, and training.
Background claim stated in paper/abstract; no experimental data provided for this statement within the abstract.
Extensive synthetic experiments show that policy regularizations reshape the narrative on what is the best DRL method for inventory management.
Paper states results from extensive synthetic experiments that change which DRL methods are considered best under policy regularization; abstract does not provide the experimental sample size, specific methods, or quantitative comparisons.
Implementation of human-replacing technologies leads to significant transformations in skill demand: it reduces reliance on low-skilled labour while increasing demand for qualified engineers, system operators and specialists in digital technologies.
Sector-specific analysis and review of international labour-market studies cited in the article documenting skill-biased effects of automation and digitalization; qualitative assessment for Ukraine's mining and metallurgical sector under workforce shortage conditions.
Foreign direct investment (FDI) shows an insignificantly positive direct effect on local TFCP but a significantly negative indirect (spillover) effect, attributed to a 'pollution haven' effect.
Spatial Durbin Model estimates for FDI on panel (30 provinces, 2010–2023): direct coefficient positive but not significant; indirect coefficient significantly negative; interpretation given as pollution-haven mechanism.
Industrial intelligence exhibits regional heterogeneity: a significantly negative direct effect in the east, a significantly positive direct effect in the central region, an insignificant direct effect in the west, and positive indirect (spillover) effects in the east and west.
Regional/subsample Spatial Durbin Model analyses dividing the sample into east, central, and west regions (30 provinces, 2010–2023); reported region-specific direct and indirect coefficients and significance levels.
Industrial intelligence has an insignificantly negative direct effect on local TFCP, but its positive spatial spillover effect is significant at the 1% level, producing a significantly positive total effect.
Spatial Durbin Model results for industrial intelligence on panel (30 provinces, 2010–2023): direct coefficient negative and not statistically significant; indirect coefficient positive and significant at 1%; total effect positive and significant.
China's TFCP rose overall from 2010 to 2023 but exhibited a widening regional gap of 'higher in the east, lower in the west'.
Panel data of 30 Chinese provincial-level regions (2010–2023); TFCP measured using an undesirable-output super-efficiency SBM model and summarized temporal and spatial patterns.
The study found a significant transformation of the employment structure under the influence of artificial intelligence.
Empirical analysis using an envelope model ("input" orientation) applied to a sample of European Union countries; the paper reports modeled changes in employment structure attributable to AI diffusion.
For AI: a cohesive professional vocabulary formed rapidly in early 2024, but the practitioner population never cohered.
Empirical finding from analysis of the 8.2M resume dataset showing a rapid increase in the vocabulary-cohesion metric around early 2024 while the population-cohesion metric did not show a corresponding rise.
The framework implies threshold effects in training and capability acquisition: when the teaching horizon lies below the prerequisite depth of the target, additional instruction cannot produce successful completion of teaching; once that depth is reached, completion becomes feasible.
Model-derived threshold result described in the abstract (mathematical analysis of prerequisite depth vs. teaching horizon).
The value of information depends on whether downstream users can absorb and act on it: a signal conveys meaning only to a learner with the structural capacity to decode it (an explanation that clarifies a concept for one user may be indistinguishable from noise to another who lacks the relevant prerequisites).
Conceptual argument motivating the model; theoretical reasoning described in the paper's intro/abstract.