Evidence (7448 claims)
Adoption
5267 claims
Productivity
4560 claims
Governance
4137 claims
Human-AI Collaboration
3103 claims
Labor Markets
2506 claims
Innovation
2354 claims
Org Design
2340 claims
Skills & Training
1945 claims
Inequality
1322 claims
Evidence Matrix
Claim counts by outcome category and direction of finding.
| Outcome | Positive | Negative | Mixed | Null | Total |
|---|---|---|---|---|---|
| Other | 378 | 106 | 59 | 455 | 1007 |
| Governance & Regulation | 379 | 176 | 116 | 58 | 739 |
| Research Productivity | 240 | 96 | 34 | 294 | 668 |
| Organizational Efficiency | 370 | 82 | 63 | 35 | 553 |
| Technology Adoption Rate | 296 | 118 | 66 | 29 | 513 |
| Firm Productivity | 277 | 34 | 68 | 10 | 394 |
| AI Safety & Ethics | 117 | 177 | 44 | 24 | 364 |
| Output Quality | 244 | 61 | 23 | 26 | 354 |
| Market Structure | 107 | 123 | 85 | 14 | 334 |
| Decision Quality | 168 | 74 | 37 | 19 | 301 |
| Fiscal & Macroeconomic | 75 | 52 | 32 | 21 | 187 |
| Employment Level | 70 | 32 | 74 | 8 | 186 |
| Skill Acquisition | 89 | 32 | 39 | 9 | 169 |
| Firm Revenue | 96 | 34 | 22 | — | 152 |
| Innovation Output | 106 | 12 | 21 | 11 | 151 |
| Consumer Welfare | 70 | 30 | 37 | 7 | 144 |
| Regulatory Compliance | 52 | 61 | 13 | 3 | 129 |
| Inequality Measures | 24 | 68 | 31 | 4 | 127 |
| Task Allocation | 75 | 11 | 29 | 6 | 121 |
| Training Effectiveness | 55 | 12 | 12 | 16 | 96 |
| Error Rate | 42 | 48 | 6 | — | 96 |
| Worker Satisfaction | 45 | 32 | 11 | 6 | 94 |
| Task Completion Time | 78 | 5 | 4 | 2 | 89 |
| Wages & Compensation | 46 | 13 | 19 | 5 | 83 |
| Team Performance | 44 | 9 | 15 | 7 | 76 |
| Hiring & Recruitment | 39 | 4 | 6 | 3 | 52 |
| Automation Exposure | 18 | 17 | 9 | 5 | 50 |
| Job Displacement | 5 | 31 | 12 | — | 48 |
| Social Protection | 21 | 10 | 6 | 2 | 39 |
| Developer Productivity | 29 | 3 | 3 | 1 | 36 |
| Worker Turnover | 10 | 12 | — | 3 | 25 |
| Skill Obsolescence | 3 | 19 | 2 | — | 24 |
| Creative Output | 15 | 5 | 3 | 1 | 24 |
| Labor Share of Income | 10 | 4 | 9 | — | 23 |
A phased adoption approach (assess needs → pilot low-risk use cases → scale modularly) is recommended to reduce risk and improve outcomes for SME AI projects.
Synthesis of best-practice guidance and pragmatic recommendations from case studies and policy literature; not empirically validated as a universal causal strategy in LMIC SMEs within the review.
External market pressures and customer demand often drive AI adoption decisions in SMEs.
Surveys and market analyses from the literature indicating demand-side pressures as adoption triggers; evidence mainly observational.
Access to finance, including scalable and blended financing models, is a key enabler for SME AI adoption.
Policy reports, case studies and financial analyses discussed in the review that identify financing availability and instrument design as central constraints/enablers; evidence is descriptive and context-dependent.
Local innovation ecosystems (universities, incubators, private-sector partnerships) support SME uptake of AI.
Case studies and ecosystem analyses in the reviewed literature documenting successful university–industry linkages and incubator support facilitating technology transfer and skills development.
Supportive government policy and adaptive regulation are important enablers of AI adoption among SMEs.
Synthesis of policy reports and governance literature included in the review identifying regulatory clarity and supportive policy as common enabling factors.
AI can improve market access for SMEs (e.g., via digital platforms and AI-enabled credit scoring) and enable potential value-chain upgrading.
Policy analyses and case-study evidence showing digital platforms and algorithmic credit assessment opening opportunities for SMEs; examples referenced from Botswana and similar LMIC contexts.
AI adoption supports new product/service innovation and faster time-to-market for SMEs.
Qualitative case studies and practitioner reports cited in the review showing instances of AI assisting R&D, prototyping, and launch processes; limited systematic quantitative measurement across sectors.
AI-enabled customer segmentation and personalization can increase sales and customer retention for SMEs.
Empirical examples and case studies from the literature and policy reports documenting improved targeting and retention in firms that adopted AI tools; evidence is largely observational and context-specific.
AI can generate productivity gains for SMEs through automation and process optimization.
Multiple case studies and firm-level surveys reported in the literature showing examples of automation-related efficiency improvements; no large-scale randomized or causal studies cited that uniformly quantify effect sizes across LMIC SMEs.
Anticipatory analytics and automated decision support can improve public resource allocation and reduce response lag, raising public sector productivity and potentially changing demand for private sector services.
Aggregate claims from empirical cases and theoretical pieces in the review that report or argue for efficiency/productivity gains from predictive systems; synthesis across several studies in the 103‑item corpus.
Realizing economic and social benefits from public‑sector AI requires interoperable, ethical‑by‑design systems combined with sustained investments in skills, infrastructure, and accountability mechanisms.
Prescriptive synthesis from the systematic review that aggregates recommendations across empirical studies and institutional reports within the 103‑item corpus.
Big Data and AI are enabling a shift in public governance from reactive to anticipatory decision-making and resource allocation.
Synthesis from a PRISMA-guided systematic review of 103 peer‑reviewed articles and institutional reports (2010–2024) mapping empirical cases of predictive analytics and AI deployment in public-sector domains.
RAG approaches (cloud or on-prem) outperform a zero-shot baseline (base model without retrieval) on retrieval/generation performance.
Empirical comparative experiments included a zero-shot base model baseline, GPT RAG cloud, and on-prem RAG; summary implies comparative superiority of RAG over zero-shot but does not provide exact metrics or sample sizes.
On-prem solutions simplify compliance with data sovereignty and privacy regulations (e.g., GDPR) and reduce legal risk for firms handling sensitive IP.
Policy-relevant assessment in environment/security evaluation arguing on-prem architectures ease regulatory compliance; no legal-case study evidence provided in summary.
Converting variable token/API costs into fixed on-prem costs can lower marginal cost per query for sustained, high-volume usage typical of some SMEs.
Economic/cost-structure analysis in the paper arguing that capex + ops converts variable to fixed costs and reduces marginal cost per query for sustained usage; no numeric break-even analyses reported in summary.
On-prem deployment materially improves data sovereignty and reduces risk of external data leakage.
Environment/security evaluations including threat/surface analysis and policy assessment arguing that on-prem architectures prevent external transmission of sensitive data; no empirical breach incidence data provided.
On-Premise RAG eliminates recurring token/API costs associated with cloud LLMs, reducing long-run OPEX.
Organizational cost accounting comparison between recurring cloud/API expenses and on-prem capital and operational costs presented in the TOE-grounded analysis; no dollar amounts or time horizons reported in summary.
On-Premise RAG outperforms commercial RAG on qualitative dimensions (usefulness and relevance) in specialized manufacturing domains.
Human evaluation by domain experts (human-in-the-loop judgments) assessing usefulness and relevance using the on-prem pipeline with a curated knowledge base; sample size and scoring protocol not specified in summary.
Market failures—data externalities, coordination failures, and large fixed costs for sensorization/computing—likely lead to underinvestment by private actors and justify targeted public interventions (data platforms, co-financing, standards).
Economic reasoning informed by observed underinvestment patterns in investment datasets and the structure of costs for sensorization/computing; institutional review indicating coordination gaps.
Institutional determinants (data governance, standards, public infrastructure) materially influence AI diffusion and should be incorporated explicitly into diffusion models alongside human capital and capital-cost channels.
Cross-country trend comparisons and institutional analysis demonstrating correlations between institutional variables and adoption/diffusion patterns; theoretical synthesis.
Workers are increasingly treating AI adoption as a collective bargaining and political issue, using strikes, bargaining demands, and internal organizing to contest deployments.
Synthesis of reports, case studies and contributions to the AIPOWW symposium documenting worker organizing episodes and demands related to AI deployments; no systematic dataset or sample size reported.
Policy recommendations include investing in workforce reskilling, promoting interoperability and data portability, designing proportional risk-based regulation, using regulatory sandboxes and staged deployment, and supporting capacity building for low- and middle-income countries to avoid an AI divide.
Synthesis of policy analysis, sectoral findings and normative recommendations derived from the comparative review and gap analysis.
AI adoption can raise firm- and sector-level productivity, potentially lifting aggregate output; measuring AI’s contribution requires new indicators of 'AI intensity'.
Economic reasoning and review of literature; recommendation for measurement approaches (software/hardware investment, AI talent, use of AI services). No primary empirical measurement provided.
Regulatory design should be context-sensitive and ethics-grounded rather than one-size-fits-all.
Normative evaluation and synthesis of governance frameworks and identified gaps across jurisdictions; policy recommendations grounded in ethical principles (transparency, fairness, accountability, human rights).
AI capabilities (learning, reasoning, perception, NLP) are being integrated rapidly across healthcare, finance, education, transportation, security and justice, producing major efficiency and service-quality gains.
Sectoral case studies and documented examples cited in policy/regulatory texts and secondary literature; comparative analysis of deployments across the listed sectors.
AI is driving large productivity and capability gains across sectors.
Synthesis of sectoral case studies and secondary literature across healthcare, finance, education, transportation, security and justice; comparative policy and regulatory analysis of documented AI deployments. No large-scale primary quantitative impact evaluation reported.
Environmental-performance labeling and user opt-outs could create demand for 'eco-optimized' models and influence competition among providers.
Market analysis in implications section (theoretical consumer preference/differentiation effects).
Mandatory inference benchmarks and public reporting would create market and regulatory incentives to optimize models for energy efficiency (e.g., compression, routing, edge inference).
Policy implications / market design analysis describing likely provider responses to benchmarking and public reporting.
Mandatory model-level disclosure and user-choice rights would help internalize negative environmental externalities, shifting costs into firms’ deployment and pricing decisions.
Economic-policy analysis in the implications section (conceptual/incentive reasoning based on disclosure->price/internalization mechanisms).
The paper recommends international coordination to prevent regulatory arbitrage and ensure consistent standards for model-level environmental governance.
Policy design and cross-jurisdictional analysis arguing for harmonization to avoid compute relocation/obfuscation and regulatory gaps.
Investors and regional planners can use the Hub to identify emerging opportunity hubs and prioritize economic development or infrastructure to support skill formation.
Implications and use-case examples in the paper proposing the Hub's application for regional strategy and investment decisions; empirical evidence for realized investment outcomes is not provided.
Policy-simulation features make it possible to compare labor-market effects of alternative interventions (subsidies, regulations, training programs) before deployment.
Description of policy simulation dashboards and scenario-analysis capabilities in Methods and Implications sections; no quantitative validation details provided in the summary.
Geospatial hotspot identification enables region-specific training investments and curricula alignment with projected demand.
Implications section connects geospatial hotspot outputs to targeted reskilling/education policy; empirical effectiveness of doing this is implied by experimental claims but not quantitatively substantiated in the summary.
The Hub supports more targeted, data-driven workforce and policy decisions by producing actionable, interpretable outputs and scenario comparisons.
Paper's Main Finding and Implications sections arguing that outputs enable targeted reskilling, policy design, and regional strategy. Empirical support is claimed via an experimental evaluation but detailed results are not reported in the summary.
Experimental evaluation shows the Hub can quantify how automation and policy interventions alter future workforce readiness.
Paper describes scenario analysis and reports that the system quantifies impacts of automation and policy in experiments, but does not provide numeric results, evaluation methodology, or datasets in the provided summary.
Experimental evaluation shows the platform can pinpoint high-potential regional opportunity hubs.
Paper claims experimental results demonstrate ability to highlight regional opportunity hubs; evaluation details (data sources, sample size, metrics) are not provided in the summary.
Experimental evaluation shows the system can identify critical talent shortages.
Paper reports an experimental evaluation that the platform can surface critical shortages; no datasets, sample sizes, numerical metrics, or evaluation design details are reported in the abstract/summary.
Hybrid approaches may deliver the best economic return by reducing need for large-scale primary data collection while maintaining acceptable performance, but they require modest real-data collection costs for fine-tuning and validation.
Inferred from comparative evaluations and economic reasoning in the reviewed literature that contrast synthetic-only, real-only, and hybrid strategies; evidence is suggestive rather than pooled quantitative analysis.
Hybrid datasets (synthetic data combined with real patient data) consistently yield better model performance than synthetic-only training across reviewed studies.
Critical literature review and thematic synthesis of machine-learning evaluation studies reported in peer-reviewed articles, technical reports and policy analyses across searched databases (Scopus, Web of Science, PubMed, Google Scholar). The review reports a recurring pattern across multiple studies, though the number of studies and exact effect sizes are heterogeneous and not enumerated in the paper; limitations noted include publication bias and heterogeneity in outcome metrics.
International certification protocols tied to explainability and safety standards would influence investment incentives and market structure.
Policy and economic analyses in the literature synthesis arguing how standards/certification shape firm behavior and investment; no empirical causal estimation provided.
A tiered risk-management framework that allocates governance intensity to interventions by clinical criticality and autonomy is recommended to maximize benefits while containing harms.
Authors' policy recommendation derived from literature synthesis of governance frameworks, risk analyses, and implementation studies; prescriptive rather than empirically validated in large-scale trials.
Federated learning and privacy-preserving collaboration can combine data advantages without centralizing sensitive records and may reduce duplicated validation costs over time.
Technical literature and pilot studies on federated learning and privacy-preserving methods summarized in the paper; limited large-scale, long-term deployment evidence noted.
Centralized updates and monitoring by vendors can reduce operational burden for healthcare providers.
Comparative analyses and deployment reports contrasting vendor-managed services with self-managed open-source deployments; synthesized evidence and stakeholder commentary.
Open-source models enable customization and local retraining that can align models with institutional workflows and patient populations.
Cross-disciplinary literature synthesis and case reports describing local retraining/customization practices; comparative analyses of model adaptability. Evidence is drawn from diverse deployments rather than controlled trials.
Clear, harmonized regulation and procurement strategies can stimulate domestic AI suppliers, reduce dependency on foreign vendors, and capture more local economic value.
Policy analysis and market-structure discussion in the review, supported by international comparisons (e.g., Singapore, EU) and procurement case studies cited among supplementary documents.
Prioritizing AI for primary care and diagnostic applications can yield high-value health returns (reduced morbidity, earlier treatment) and improve system efficiency.
Synthesis of clinical application studies and health-economics literature within the 2020–2025 review timeframe; specific quantified returns were not uniformly reported across primary sources in the summary.
Public investment in digital health infrastructure (broadband, cloud/edge compute, interoperable data systems) is a precondition for scalable returns from AI; underinvestment will dampen both health and economic gains.
Economic and systems analysis presented in the review, drawing on international benchmarking and health-economics literature; arguments are analytical and based on modeled or literature-supported relationships rather than specified local experimental data.
AI for diabetic retinopathy screening reported an accuracy of approximately 89.3% in reviewed studies.
Reported summary statistic drawn from diagnostic performance studies identified in the 2020–2025 literature review; exact primary study sample sizes and study designs not provided in the summary.
Indonesia has demonstrated strong clinical efficacy of AI in healthcare, notably in diagnostics, telemedicine, and chronic disease management.
Narrative synthesis of literature (2020–2025) and thematic analysis of studies and pilot programs included in the review; sources include PubMed, Google Scholar, Garuda, SINTA, and 42 supplementary documents (national policy papers, SATUSEHAT governance reports, Delphi consensus studies). Specific primary study details (sample sizes, study designs) vary by application and are not uniformly reported in the synthesis.
There is a need for standards on provenance, licensing, and security auditing of AI-generated code, and potential roles for certification and liability frameworks.
Policy recommendation grounded in the identified IP, licensing, and security gaps from the literature synthesis.