Evidence (9875 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
Filtered →
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 |
Adoption
Remove filter
Transparent, auditable AI systems and governance mechanisms are necessary to maintain public trust and democratic oversight.
Normative and governance-focused argument in the book; supported by conceptual reasoning rather than empirical public-opinion or audit studies in the blurb.
Designing AI systems with participation and accessibility at their core is essential to prevent concentration of gains and widening inequalities.
Normative recommendation based on equity concerns and policy analysis; not empirically tested or quantified in the blurb.
AI platforms can materially improve efficiency and resilience of supply chains, altering comparative advantage and regional integration dynamics.
Illustrative vignette (logistics optimization) and policy-analytic reasoning; no empirical supply-chain studies or measured efficiency gains reported in the blurb.
Labor-market policy should emphasize reskilling, algorithmic job-matching, and social safety nets to account for rapid compositional changes enabled by AI platforms.
Policy recommendation grounded in scenario analysis and applied-AI descriptions; no empirical evaluation or quantified labor market impact provided in the blurb.
Policymakers need new institutional capacities to integrate AI-driven foresight into fiscal, trade, and labor policymaking.
Policy analysis and prescriptive argument in the book; illustrated with scenario reasoning but lacking empirical measurement of capacity gaps or interventions.
Rather than replacing human judgment, AI augments foresight and adaptation, enabling resilient, inclusive, and participatory governance if guided by deliberate policy design.
Normative and conceptual argumentation with illustrative vignettes (e.g., policymaker vignette); no empirical validation or sample sizes reported.
AI is transforming economic decision-making, governance, and value creation across sectors and countries.
Conceptual synthesis presented in the book/blurb; no empirical study or sample reported—claim supported by cross-sector examples and narrative argumentation.
Policy interventions—investments in digital infrastructure, vocational and continuing education, and incentives for firm-level training—amplify AI benefits, particularly in lower-income countries.
Policy-relevant heterogeneous treatment effects and simulated counterfactuals showing larger productivity gains in contexts with better infrastructure and training; empirical interaction terms between policy proxies and adoption effects.
Cross-country differences in AI effects are driven by digital infrastructure, human capital, and the regulatory environment.
Regression analyses interacting AI adoption with country-level indicators (broadband penetration, tertiary education rates, regulatory indices) and observing systematic variation in estimated productivity impacts.
Productivity improvements from AI spill over to upstream suppliers in the same value chain.
Input-output linked firm analyses and supplier-customer matched panels showing productivity increases among upstream firms when downstream partners adopt AI; event-study timing consistent with spillovers.
AI benefits are greatest where AI adoption is combined with worker training, cloud infrastructure, and managerial changes (complementarity effect).
Interaction analyses in firm-level regressions and stratified comparisons showing larger productivity gains for adopters that also report training programs, cloud adoption, or management practices; robustness checks controlling for firm fixed effects.
High-income countries experience larger productivity gains from AI (roughly 8–12%) and faster reallocation toward higher-skilled tasks.
Heterogeneity analysis using country-level indicators (income classification, tertiary education rates) and worker-level linked employer-employee microdata; interaction terms in difference-in-differences and occupation-level event studies.
Firms using advanced AI report a 5–12% increase in measured labor productivity within 1–3 years after adoption (average effect).
Panel estimates from multiple country firm-level datasets using difference-in-differences and event-study specifications with 1–3 year post-adoption windows and controls/robustness checks to bound potential selection.
A certification/audit industry is likely to emerge (market for algorithm auditors, explainability tools, compliance software).
Market-outcome inference in the economics implications section; forecast based on anticipated demand for compliance/audit services following white‑box mandates.
The protocol projects and systematizes 16 anticipated constitutional rulings by the SCJN to create enforceable standards.
Legal-methodological approach described in the compendium: explicit projection and systematization of 16 anticipated SCJN rulings to derive standards.
Greater transparency and audit trails improve regulators’ ability to monitor concentration risks, model commonality and systemic vulnerabilities arising from algorithmic homogenization.
Policy analysis and regulatory design argument in the compendium, drawing on macroprudential principles and comparisons with European regulatory approaches; not empirically tested within the paper.
Regulatory certainty around rights‑based standards may reorient investment toward explainable AI, compliance tooling, audit services and governance technologies — creating a potential new sector of AI‑economics activity.
Projection based on market response theory and industry trends noted in the compendium; supported by comparative regulatory cases but not by quantified investment data in the paper.
Localized datasets and mandated disclosure could create public datasets and benchmarks that improve model fairness and enable new entrants.
Policy design proposal and comparative precedent examples in the corpus; normative expectation rather than demonstrated outcome.
Transparency standards can reduce information asymmetries between firms, borrowers and regulators, potentially lowering adverse‑selection problems in lending markets.
Theoretical economic argument grounded in market microstructure and information economics; supported by comparative regulatory literature in the corpus (no new empirical estimation reported).
Non‑discrimination and fairness requirements (procedural standards and substantive tests) must be mandated to prevent biased exclusion in automated credit and financial services.
Doctrinal analysis of jurisprudence and regulatory materials, comparative law review (Mexico ↔ Europe), and review of technical literature on algorithmic fairness in the ~4,200‑text forensic audit.
A 'White Box' regulatory model — mandatory transparency, explainability, and forensic auditability — should be required for algorithms used in banking/fintech, particularly credit scoring.
Normative protocol design and synthesis of legal, regulatory and technical literature in the forensic audit; policy operationalization component of the compendium (method: doctrinal analysis and normative design).
Digital Sovereignty should be recognized as a fundamental human right protecting citizens’ control over algorithmic decisions affecting economic life.
Normative/doctrinal legal argumentation and comparative law synthesis across the compendium; grounded in rights‑based reasoning and alignment with international human‑rights norms (no experimental/empirical test).
The governance pattern can lower operational and integration barriers to adopting generative AI and automation, potentially accelerating diffusion across enterprises.
Theoretical and qualitative claim based on synthesis of deployment patterns and case examples; no measured adoption rates or diffusion studies provided.
AI-specific controls (testing/validation, drift detection, retraining triggers) reduce AI-related risks in enterprise automation.
Paper's prescriptive governance controls and AI risk-management recommendations based on industry practice; described qualitatively without quantitative effect sizes or controlled evaluation.
Aligning technical architecture with organizational governance structures (roles, approval workflows, risk committees) and following a lifecycle (design → validation → deployment → monitoring → decommissioning) is necessary for operationalizing automation governance.
Cross-case lessons and organizational integration recommendations derived from multi-sector case examples and best-practice synthesis; presented as prescriptive architecture and lifecycle processes.
Embedded governance features (access/data usage policy enforcement, model-output controls), human-in-the-loop checkpoints for high-risk decisions, continuous monitoring, and audit trails increase accountability and provide regulatory evidence.
Normative recommendations grounded in industry best practices and case examples; pattern specification enumerating governance controls. Evidence is qualitative rather than quantitative.
A practical reference pattern combining low-code development, RPA, generative AI, and a centralized governance layer can be deployed in mission-critical ERP/CRM landscapes.
Architectural pattern design and cross-case lessons from multi-sector enterprise implementations; qualitative synthesis of industry best practices and case examples. No large-scale quantitative deployment statistics provided.
Embedding policy enforcement, risk controls, human oversight, and continuous monitoring into the automation lifecycle enables organizations to scale automation while preserving data protection, regulatory compliance, operational stability, and long-term system integrity.
Conceptual framework synthesized from industry best practices and comparative analysis of multi-sector enterprise implementations and case examples; architectural pattern design. Methods: qualitative synthesis and pattern extraction. No randomized or large-sample empirical evaluation reported.
Design choices that combine transparency and explainable personalization materially increase consumer trust and purchase intention, making them important levers for firms seeking higher conversion in AI-mediated commerce.
Inference drawn from experimental findings showing transparency and empathetic personalization increased trust (and via trust, purchase intention); applied as an implication for firms.
Higher digital literacy weakens (attenuates) the negative link from perceived manipulation to purchase intention.
Moderator analysis in PLS-SEM including measured digital literacy as a moderator of the perceived manipulation → purchase intention path in the experimental sample (UAE, ages 18–25).
Trust is the primary (dominant) mediator through which transparency and empathetic personalization increase purchase intention.
Mediation analysis within PLS-SEM on experimental data (2 × 2 design); measures include trust and purchase intention; indirect paths from design cues to purchase intention were analyzed.
An empathetic, personalized conversational tone in chatbots increases trust among young consumers (UAE, ages 18–25).
2 × 2 between-subjects experiment manipulating conversational tone (empathetic/personalized vs. generic), same sample (UAE, ages 18–25); trust measured; analyzed with PLS-SEM.
Transparent AI identity disclosure increases trust among young consumers (UAE, ages 18–25).
2 × 2 between-subjects experiment manipulating identity disclosure (AI transparent vs. nondisclosed), sample: young consumers in the UAE aged 18–25; trust measured as a dependent variable; effects estimated using PLS-SEM.
Effective regulation can reshape market equilibria by mandating transparency/audits, enabling interoperability/identity portability, constraining high-risk personalization practices, and requiring privacy-preserving measurement standards.
Policy and economic modeling arguments combined with case examples; prescriptive claim based on plausibility and prior regulatory impacts rather than new causal estimates.
Regulatory interventions (e.g., limits on third-party cookies or profiling) will redirect long-term investments toward privacy-preserving measurement and contextual advertising solutions.
Policy analysis and plausibility argument based on past regulatory changes (cookie deprecation) and industry responses; predictive, not empirically validated within the paper.
Improvements in targeting raise advertiser willingness-to-pay, shifting surplus toward platforms unless competitive pressures or regulation change fee structures.
Economic theory and observed industry trends; no new cross-sectional or panel data regression in this paper to quantify the shift.
Interpretable models, causal evaluation of impact (not only prediction metrics), privacy-by-design, and governance mechanisms are central to sustainable adoption (resilience criteria).
Recommended evaluation framework based on methodological critique (attribution complexity, metric misalignment) and best-practice literature; no empirical validation sample provided.
Long-run viability requires moving beyond raw predictive performance toward resilient, interpretable, policy-aware, and socially legitimate systems.
Normative recommendation grounded in evaluation challenges and literature on trustworthy AI; not an empirically tested hypothesis within the paper.
Regulation shapes incentives for architectures (e.g., favoring first-party data architectures over third-party tracking) (Innovation vs regulatory compliance trade-off).
Policy analysis and observations about industry responses to cookie deprecation and privacy regulation; descriptive industry trend evidence rather than a single empirical trial.
Complementarities matter: digitalization increases AGTFP more when combined with complementary investments and institutions (mechanization, R&D, cooperative organization).
Findings from mediation analysis and interaction/heterogeneity checks indicating larger effects where complementary inputs/institutions are present.
Non-grain-producing provinces experience larger AGTFP gains from digital rural development than major grain-producing provinces.
Comparative sub-sample analysis (non-grain vs. major grain-producing regions) showing larger estimated effects in non-grain-producing areas.
Digital service capacity shows diminishing marginal returns: the marginal positive effect of digital services on AGTFP weakens at more advanced stages of digital-service development.
Panel threshold/modeling of nonlinearity indicating a decreasing marginal effect of the digital service sub-index on AGTFP at higher development levels.
Digitalization accelerates agricultural mechanization and the diffusion of agricultural R&D, which act as channels raising AGTFP.
Mediation analysis including mechanization rate and agricultural R&D input/technology diffusion indicators as mediators; reported significant indirect effects.
Digital rural development strengthens cooperative organizational forms (farmer cooperatives), and this organizational upgrading contributes to higher AGTFP.
Mediation tests showing digitalization is associated with greater cooperative organization indicators, which in turn are associated with higher AGTFP.
Digital rural development encourages larger-scale agricultural operations (land consolidation/scale expansion), which contributes to increases in AGTFP.
Mediation models that include farm scale/land transfer indicators as mediators and find significant indirect effects; analysis notes institutional constraints limit full realization.
Digital rural development raises AGTFP in part by promoting labor mobility and reallocating labor toward higher-productivity uses.
Mediation analysis using the same provincial panel (2012–2022) showing significant indirect effects through labor reallocation/factor allocation variables.
Empirical models of labor costs, productivity, and AI adoption should use total labor cost (wages + NWC) rather than wages alone; CFIL should be included when modeling transitions from informal to formal employment under automation scenarios.
Methodological recommendation based on the magnitude of measured non-wage and formalization costs (2023 estimates for 19 countries) and implications for correctly specifying empirical models; not an empirical test but a suggested best practice.
Macroeconomic and fiscal gains (GDP growth and increased tax revenues) from platform-enabled productivity are quantitatively estimated via input–output/CGE-style simulations but remain sensitive to assumptions about adoption and policy.
Computed economy-wide estimates from input–output or computable general equilibrium simulations that scale micro productivity improvements; sensitivity analyses run under alternative adoption and policy scenarios.
Observed productivity and participation effects are attributable to AI-enabled capabilities using comparative or quasi-experimental contrasts (e.g., before/after rollouts, adopter vs non-adopter, geographic variation in fulfillment infrastructure).
Identification strategy described: comparative/quasi-experimental contrasts across time, sellers, and geographies; robustness and sensitivity checks reported to support causal attribution.
Algorithmic advertising, dynamic pricing, and demand-forecasting measurably improve ad-targeting outcomes and pricing responsiveness, increasing listing conversions and sales for adopting sellers.
Demand-side algorithmic performance measures (ad-targeting precision/CTR, conversion rates before/after dynamic pricing adoption) and seller sales metrics from platform data and quasi-experimental contrasts.