Evidence (14922 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
9047 claims
Filter claims →
Productivity
8066 claims
Filter claims →
Governance
7278 claims
Filter claims →
Human-AI Collaboration
6912 claims
Filter claims →
Org Design
4439 claims
Filter claims →
Innovation
4359 claims
Filter claims →
Labor Markets
3652 claims
Filter claims →
Skills & Training
3018 claims
Filter claims →
Inequality
2160 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 | 795 | 210 | 105 | 955 | 2131 |
| Governance & Regulation | 886 | 414 | 197 | 126 | 1654 |
| Organizational Efficiency | 826 | 204 | 129 | 87 | 1257 |
| Technology Adoption Rate | 681 | 259 | 128 | 110 | 1189 |
| Research Productivity | 464 | 138 | 65 | 349 | 1028 |
| Output Quality | 503 | 196 | 61 | 53 | 813 |
| Decision Quality | 351 | 180 | 84 | 51 | 673 |
| AI Safety & Ethics | 238 | 288 | 71 | 34 | 637 |
| Firm Productivity | 455 | 58 | 92 | 20 | 631 |
| Market Structure | 186 | 172 | 123 | 25 | 511 |
| Task Allocation | 222 | 70 | 76 | 34 | 407 |
| Innovation Output | 238 | 28 | 48 | 18 | 334 |
| Skill Acquisition | 177 | 62 | 62 | 17 | 318 |
| Employment Level | 107 | 57 | 108 | 13 | 287 |
| Fiscal & Macroeconomic | 135 | 72 | 44 | 26 | 284 |
| Firm Revenue | 172 | 50 | 28 | 5 | 256 |
| Consumer Welfare | 121 | 68 | 45 | 12 | 246 |
| Task Completion Time | 183 | 33 | 10 | 13 | 240 |
| Inequality Measures | 45 | 126 | 50 | 6 | 227 |
| Worker Satisfaction | 95 | 74 | 23 | 12 | 204 |
| Error Rate | 77 | 98 | 11 | 4 | 190 |
| Regulatory Compliance | 84 | 73 | 17 | 7 | 181 |
| Automation Exposure | 61 | 61 | 27 | 14 | 166 |
| Training Effectiveness | 98 | 21 | 14 | 19 | 154 |
| Wages & Compensation | 78 | 37 | 25 | 6 | 146 |
| Developer Productivity | 105 | 18 | 14 | 6 | 144 |
| Team Performance | 87 | 17 | 28 | 10 | 143 |
| Job Displacement | 12 | 83 | 23 | 1 | 119 |
| Hiring & Recruitment | 53 | 8 | 8 | 3 | 72 |
| Social Protection | 39 | 17 | 8 | 2 | 66 |
| Creative Output | 32 | 20 | 8 | 3 | 64 |
| Skill Obsolescence | 5 | 50 | 6 | 1 | 62 |
| Labor Share of Income | 17 | 20 | 17 | — | 54 |
| Worker Turnover | 15 | 15 | — | 3 | 33 |
| Industry | — | — | — | 1 | 1 |
Perceived algorithmic standardized guidance improves food delivery riders' mental health by reducing work pressure.
466 Chinese food delivery riders; SEM and bootstrapping testing mediation (standardized guidance -> work pressure -> mental health) within JD-R framework.
Perceived algorithmic behavioral constraint promotes risky riding behavior among food delivery riders through increased work pressure.
Data from 466 Chinese food delivery riders; mediation tested using SEM and bootstrapping showing behavioral constraint -> work pressure -> risky riding behavior.
Perceived algorithmic tracking evaluation promotes risky riding behavior among food delivery riders through increased work pressure.
Survey data from 466 Chinese food delivery riders; SEM and bootstrapping used to test mediation (tracking evaluation -> work pressure -> risky riding behavior).
Algorithms now surpass human capability in processing speed, pattern recognition and data-driven decision-making.
Asserted in the paper's opening claims as a general factual premise; grounded in the paper's literature grounding but no original empirical tests or sample reported.
Research on large language models (LLMs) has increased especially after the release of ChatGPT.
Temporal/topic-prevalence analysis in the corpus indicating a rise in LLM-related topic weight following the ChatGPT release date.
There is significant research concentration on AI applications in supply chains, labor markets, and large language models (LLMs).
Topic-modeling results showing relatively high prevalence of topics labeled as supply chains, labor markets, and LLMs in the >4,600-paper corpus.
Education, reskilling, and institutional responses are important in shaping the economic outcomes of artificial intelligence.
Policy implication derived from the observed/modeled heterogenous effects of AI on occupations and productivity; presented as a normative recommendation rather than an empirically tested result in the provided text.
Productivity gains associated with AI may support long-term economic growth.
Reference to productivity data and growth theory linking productivity improvements to long-run growth; the paper states this as a potential outcome but does not provide quantified long-run estimates or empirical identification in the excerpt.
AI complements higher-skill labor.
Interpretation of labor market data patterns and theoretical task-complementarity arguments presented in the paper; empirical details (which datasets, estimation strategy, sample size) are not provided in the text excerpt.
Artificial intelligence is a skill-biased technological innovation.
Framing and argumentation in the paper situating AI within the skill-biased technical change literature; references to analyses of publicly available labor market and productivity data (sources, time periods, and sample sizes not specified in the text).
Firms' technical competencies amplify the positive effect of AI adoption on performance.
Moderation analysis in the PLS-SEM using the same 280-SME survey indicating a significant positive moderating role for technical/technical competency measures.
Firms' financial capacity amplifies the positive effect of AI adoption on performance.
Moderation analysis within the PLS-SEM on survey data from 280 Tunisian SMEs showing a significant positive moderating effect of financial strength on the AI adoption → performance link.
AI adoption significantly improves operational performance of Tunisian SMEs.
Same empirical dataset (n=280) and PLS-SEM analysis reporting a significant AI adoption → operational performance relationship.
AI adoption significantly improves financial performance of Tunisian SMEs.
Survey data from 280 Tunisian SMEs analyzed using partial least squares structural equation modeling (PLS-SEM); significance of the AI adoption → financial performance path reported in the model.
The future of AI must be guided by human-centered ethical principles, international cooperation, and strategic regulatory planning to ensure societal benefit and minimize systemic risks.
Concluding recommendation in the paper (normative/policy prescription); the abstract gives no empirical evidence or quantified analysis to demonstrate effectiveness of these measures.
Public governance is pivotal to ensuring equitable and accountable AI implementation.
Policy argument/conclusion presented in the paper; the abstract does not report empirical validation, case studies, or metrics supporting this causal claim.
Big Data Analytics and AI can improve audit accuracy and reduce costs.
Reported results from literature review and empirical analysis in the study; precise cost or accuracy metrics and sample information are not provided in the abstract.
Integrating BDA and AI within the Audit 5.0 framework represents a fundamental shift toward intelligent, adaptive, and value-driven auditing, while underscoring the need for enhanced auditor competencies and alignment with evolving regulatory and professional requirements.
Overall synthesis of literature and empirical results from the mixed-method study (systematic review + SEM-based empirical analysis in finance and technology sectors); phrased as a high-level conclusion.
There is a need for stronger governance, ethical frameworks, and targeted training to fully realize the benefits of digital auditing.
Conclusions drawn from the literature synthesis and empirical observations regarding challenges to implementing Audit 5.0; recommendation rather than a measured effect.
BDA and AI enable real-time and predictive risk assessment and enhanced fraud detection, expanding audit coverage beyond traditional sampling.
Synthesis of prior theoretical and empirical studies and the study's empirical analysis (SEM) focusing on risk assessment, anomaly detection, and continuous auditing in finance and technology sectors.
Investment in AI correlates with improved audit efficiency.
Reported empirical correlations from the study's analysis (SEM) combined with literature review; detailed metrics and sample information not included in the abstract.
Investment in AI correlates with reductions in audit restatements.
Empirical evidence cited in the study (SEM-based analysis across organizations in finance and technology); exact sample size and statistical coefficients not provided in the summary.
BDA and AI facilitate continuous auditing (real-time auditing).
Synthesis of prior literature and empirical analysis within Audit 5.0 framework; methods include systematic literature review and SEM on sectoral samples (finance and technology).
Digitalization (BDA and AI) improves audit productivity.
Empirical analysis (SEM) and literature synthesis focused on finance and technology organizations; empirical details (sample size, effect sizes) not given in the summary.
Audits supported by Big Data Analytics (BDA) and artificial intelligence (AI) significantly outperform traditional audit approaches.
Mixed-method research: systematic literature review plus empirical analysis using structural equation modeling (SEM) on organizations in the finance and technology sectors (sample size not reported in the provided text).
Workers with more sluggish beliefs remain overly optimistic in recessions, are hired at higher wages, and face a higher risk of separation.
Model results (with heterogeneous learning rates) and supporting survey patterns: slower-learning workers' beliefs lag the downturn, leading to higher accepted/hired wages in the model and higher simulated separation probabilities; the paper links these model outcomes to observed differences in transitions.
Allowing for heterogeneity in workers' learning rates explains observed differences in employment transitions.
Extended model with heterogeneous worker learning rates calibrated/validated against observed employment transition patterns (comparison of simulated transition rates to empirical patterns from survey/administrative moments).
In equilibrium, the gap between firm and worker beliefs drives unemployment volatility.
Model simulations and equilibrium analysis after calibration show that differences between firm and worker beliefs amplify unemployment volatility (simulated/quantitative result based on the calibrated model).
More optimistic workers demand higher wages.
Correlation documented in the Michigan Survey of Consumers between individual optimism about macro conditions and reported wage demands/expectations (survey-based regression evidence linking optimism to higher demanded wages).
High current usage, breadth of application, frequent use of AI tools for testing, and ease of use correlate strongly with future intended adoption.
Correlational/regression analyses of survey variables (N=147) predicting respondents' stated future intention to increase AI tool use from measures of current usage, breadth of tool applications, frequency of testing-tool use, and perceived ease-of-use.
Developers report both productivity and quality gains from using AI tools.
Aggregate self-reported responses from 147 professional developers indicating perceived improvements in productivity and code quality associated with AI tool use.
There is no perceptual support for the Quality Paradox; PP is positively correlated with Perceived Code Quality (PQ) improvement.
Statistical analysis of survey measures (N=147) showing a positive correlation between respondents' Perceived Productivity scores and their Perceived Code Quality improvement scores; absence of evidence for a negative PP–quality relationship.
Frequent and broad AI tools use are the strongest correlates of both Perceived Productivity (PP) and quality, with frequency strongest.
Correlational analysis of self-reported survey responses from a sample of 147 professional developers measuring AI tool usage frequency and breadth and perceived outcomes (Perceived Productivity and Perceived Code Quality).
AI-influenced efficiency has a statistically significant but moderate positive impact on reducing the oil and gas trade deficit and on GDP growth.
Quantitative macroeconomic assessment (second hypothesis) reported in the paper indicating statistically significant, albeit moderate, positive effects of AI-driven efficiency on macro indicators (GDP growth and oil & gas trade balance).
AI-based real-time optimization in fuel blending surpasses traditional modeling approaches and reduces waste.
Comparative model validation and application evidence reported alongside the R2 = 0.99 result; qualitative statements that AI optimization outperforms traditional models and reduces material waste.
Predictive maintenance (PdM) systems powered by advanced AI methods ensure continuous operation and extend the life of critical hydrocarbon assets.
Qualitative and case-based evidence from described AI applications in the downstream sector within the mixed-methods study (examples of PdM deployments and reported operational outcomes).
AI adoption in the downstream petroleum sector is significantly positively correlated with improved operational efficiency.
Quantitative analysis within the mixed-methods study assessing immediate impact of AI on downstream operational efficiency (first hypothesis); reported as a statistically significant positive correlation (method described as quantitative assessment of operational metrics following AI adoption).
Through a comparative analysis of pioneering AI strategies in Rwanda, the United Kingdom, the United States, China, and Australia, this paper demonstrates how the DARE framework can serve as both a diagnostic tool to identify national gaps and a prescriptive blueprint for building a more equitable, human-centric automated future.
Reported method in abstract: comparative analysis of five countries (Rwanda, UK, US, China, Australia). The abstract claims demonstration but does not detail the analytic method, metrics, or sample beyond the five-country comparison.
AI promises unprecedented productivity gains.
Asserted in abstract; no empirical evidence or quantification provided in the abstract.
Given current evidence, there is greater scope for task reconfiguration and augmentation in exposed occupations than for immediate large-scale displacement.
Synthesis of task-level capability mapping and occupational complementarity analysis showing that many exposed tasks are complementary (augmentable) rather than directly substitutable, and firm-level adoption evidence showing limited job losses to date.
Most jobs that are exposed to AI in the Philippines also exhibit high complementarity with AI, suggesting substantial scope for augmentation rather than immediate displacement.
Complementarity analysis using Philippine labor force data (task- and occupation-level measures of complementarities) together with task-level evidence on what generative AI can perform in practice.
ERM adoption is associated with stronger organizational resilience during crises (for example, global pandemics).
Empirical studies from the reviewed literature—including studies covering crisis periods—report associations between ERM practices and resilience; specific study designs and sample sizes vary and are not detailed in the summary.
ERM adoption improves MSMEs' access to external financing.
Reviewed empirical evidence reported in the article that links ERM adoption to enhanced external financing outcomes for firms; details of individual studies (methods, n) are not provided in the summary.
ERM implementation is associated with sales growth and revenue stability for MSMEs.
Aggregate findings from empirical studies included in the literature review indicating links between ERM adoption and sales/revenue outcomes; original sample sizes and methods vary by study and are not specified in the summary.
ERM adoption is generally associated with improved financial performance of MSMEs.
Synthesis of empirical studies reviewed in the literature review, primarily from emerging economies; specific study designs and sample sizes vary across the reviewed literature (not reported in the summary).
Adopting a standardised yet flexible approach to incentive design can help produce more reliable and generalizable knowledge in human–AI decision-making research.
Authors' argument/recommendation based on their thematic review and the proposed framework (this is a normative claim; no empirical validation provided in excerpt).
Human judgement remains paramount for high-stakes decision-making.
Assertion in the paper framing the motivation for human–AI collaboration research (based on prior literature and domain practice; no specific empirical data or sample sizes provided in excerpt).
AI has revolutionised decision-making across various fields.
Statement in paper's introduction summarizing prior work and trends (literature-level claim; no specific studies or sample sizes provided in excerpt).
Overall, the framework improves efficiency, fairness, and quality of care in hospital workforce management.
Aggregate conclusion drawn from experiments (forecasting metrics, scheduling conflict/fairness improvements, performance evaluation results, stress tests, and pilot deployment outcomes) described in the paper.
Pilot deployments of the framework demonstrated tangible benefits, including an 18% reduction in patient waiting times and a 14% improvement in satisfaction scores.
Reported outcomes from pilot deployments (real-world trials); the number of pilot sites, duration, patient/sample sizes, and baseline comparison methodology are not detailed in the provided text.