Evidence (1945 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 |
Skills Training
Remove filter
The learned adaptive policy outperformed a fixed-wrench baseline by an average of 10.9% across five material setups.
Empirical evaluation: comparison between learned adaptive policy and a fixed-wrench policy on five different material setups; the paper reports an average improvement of ~10.9% (the exact performance metric formulation and per-setup statistics are not provided in the summary).
Integrating AI (notably ML and NLP) meaningfully automates routine software engineering tasks across requirements management, code generation, testing, and maintenance.
Systematic literature review of prior AI-for-SE work combined with an empirical survey of software engineering professionals reporting usage and examples of tool-supported automation; sample size for the survey not specified in the summary.
Policy recommendations include subsidizing complementary investments (data governance, training) rather than technology-only incentives; encouraging standards and interoperability; and funding evaluation studies to measure distributional effects and long-run productivity impacts.
Authors' policy section proposing these interventions based on case findings and broader policy implications.
The authors propose a conceptual optimisation framework emphasizing three pillars: digital integration (tech stack & data), collaboration (processes & governance), and continuous improvement (metrics, feedback loops).
Paper presents a conceptual framework derived from cross-case findings; theoretical/conceptual contribution rather than empirical estimation.
Explanations must be tailored to stakeholders (clinicians, regulators, customers) and integrated into decision processes to be useful (human-centered design principle).
Thematic coding of design and HCI literature within the review; draws on empirical studies and design guidance recommending stakeholder-specific explanation formats and integration into decision workflows.
Standards and governance frameworks (for model auditability, security, and alignment) will become economic infrastructure influencing adoption costs and market trust.
Conceptual argument linking governance to adoption and trust, drawing on normative risk analysis; no empirical governance impact studies included.
Increasing AI autonomy magnifies ethical, safety, and value‑alignment concerns; robust human oversight and institutional governance are required.
Normative and risk analysis based on projected increases in system autonomy and illustrative failure modes; no formal safety audits included.
A one standard-deviation increase in AI adoption causally increases employment in occupations requiring complex problem-solving and interpersonal skills by 1.8%.
Same panel (38 OECD countries, 2019–2025) and AI Adoption Index; IV estimation with occupational employment classified by task type (complex problem-solving & interpersonal); fixed effects and robustness checks reported.
The paper proposes a research agenda prioritizing interoperable, ethical‑by‑design platforms; metrics to measure social equity impacts; and adaptation of global standards to local institutional capacities.
Explicit list of three prioritized research directions provided in the paper, derived from the systematic synthesis of the 103 items.
High‑income examples (e.g., Estonia, Singapore) demonstrate mature integration of digital/AI systems in e‑government, urban mobility, and e‑health.
Empirical case examples drawn from the reviewed literature and institutional reports cited in the review; specific country examples (Estonia, Singapore) repeatedly referenced as mature adopters.
The paper introduces a Predictive Skill Gap Intelligence Hub — an AI-driven platform that combines macro- and micro-level indicators with probabilistic growth models and intelligent skill-synthesis to proactively forecast regional and sectoral labor demand–supply gaps.
Description of system architecture and modeling approach in the paper (methods section). No numerical evaluation metrics or datasets provided for this descriptive claim.
Coordinated policy reform, targeted infrastructure investment, workforce training, and equity-focused implementation are strategic priorities to realize AI’s potential in Indonesian healthcare.
Consensus recommendations drawn from the narrative synthesis, thematic analysis, and Delphi consensus studies included among the 42 supplementary documents and the broader 2020–2025 literature body.
Recommended research priorities for economists include measuring how adoption changes task mixes and wages, quantifying verification/remediation costs, estimating productivity gains net of security/IP costs, and studying market dynamics from centralized model providers.
Author recommendations based on identified gaps in the empirical literature synthesized by the paper.
Cognitive interlocks include concrete mechanisms such as policy-enforced gates, automated verification thresholds, role-based checks, and mandatory rebuttal workflows to force verification before outputs are trusted or deployed.
Design details and enumerated mechanisms within the Overton Framework as presented in the paper; no implementation case studies reported.
The Overton Framework is an architectural remedy that embeds 'cognitive interlocks' into development environments to enforce verification boundaries and restore system integrity.
Prescriptive architectural proposal described in the paper (design specification and principles); presented conceptually without empirical validation.
Enhanced gross‑flows estimation using longitudinal microdata can better track transitions (job-to-job, upskilling, unemployment spells) and measure occupational churn and reallocation.
Established econometric practice cited in paper; recommendation to use panel/admin microdata (CPS longitudinal supplements, LEHD/LODES, UI records); no new empirical results but aligns with standard methods.
Common uses of AI among practitioners include generating code snippets, suggesting fixes, accelerating routine tasks, surfacing design patterns or documentation, and scaffolding prototypes.
Practice-focused qualitative data from interviews and workflow analysis at Netlight; authors list these use-cases as commonly reported by practitioners; frequency counts not provided.
Practitioners use AI primarily as a practical assistant (coding, debugging, prototyping, knowledge retrieval) rather than as a fully autonomous developer.
Reported practitioner accounts and observations from the Netlight field study (interviews/observations); examples of tasks AI is used for were documented in the paper; sample limited to experienced consultants at one firm.
Experienced IT professionals at Netlight are already integrating AI tools into everyday development work.
Qualitative field study conducted at Netlight Consulting GmbH using interviews, observations, and analysis of practitioner workflows; single-firm sample (Netlight); exact number of participants not reported.
Automated equivalency systems require algorithmic oversight features (audit trails, human-in-the-loop checks) to maintain trust and labor-market legitimacy.
Governance recommendation following best practices in algorithmic accountability; not supported by empirical testing of oversight mechanisms in this context.
AI tools (automated document parsing/NLP, translation, equivalency-prediction classifiers, anomaly detection) can scale credential processing and reduce transaction costs and processing time.
Paper cites potential AI capabilities and application areas; the claim is inferential from known AI functionalities, with no implementation benchmark or throughput numbers provided.
Training improved exam scores by 0.27 grade points relative to optional access without training (p = 0.027).
Intent-to-treat comparison between the optional-access-with-training arm and the optional-access-without-training arm in the randomized trial (n = 164); reported effect size = +0.27 grade points and p-value = 0.027.
A brief, targeted training increased voluntary LLM use from 26% (optional access without training) to 41% (optional access with training).
Randomized experiment with 164 law students assigned to three arms (no access, optional access, optional access + ~10-minute training). Observed adoption rates in the two optional-access arms were 26% (untrained) vs. 41% (trained).
A research agenda prioritizing empirical evaluation, model transparency, and rigorous impact assessment is required to translate conceptual promise into measurable public value.
Explicit recommendation in the blurb identifying research priorities; not an empirical claim but a proposed course of action.
Illustrative vignettes show AI in action: logistics optimization for trade, AI models for national fiscal decision-making, and algorithmic job-acceleration for individual labor market navigation.
Reference to specific case vignettes contained in the book; these are illustrative scenarios rather than empirical case studies with measured outcomes.
Ten defining policy questions structure the book’s approach, turning abstract AI capabilities into operational policy choices.
Descriptive claim about the book's organization; verifiable by inspecting the book's table of contents (no external empirical data).
International comparability in these analyses is achieved using PPP adjustments for monetary measures and standardized occupation/task classifications (ISCO/ISCO-08) with harmonized baseline years and variable definitions.
Described data harmonization procedures across multi-country firm and worker datasets, including PPP adjustments and use of ISCO classification for occupations.
Adoption of advanced AI tools (especially generative AI) raises firm-level productivity on average.
Meta-analysis of firm-level panel studies using administrative tax and manufacturing surveys and proprietary AI-usage logs; difference-in-differences and event-study estimates comparing adopters vs non-adopters with firm fixed effects and robustness checks.
There is a widespread consensus across the reviewed literature on the need for worker upskilling, active labor‑market policies, and targeted support for displaced workers.
Policy recommendations recurring in the majority of the 17 peer‑reviewed papers synthesized in the review.
Alternative training channels (self-education and professional retraining) are nontrivial contributors to the AI workforce supply.
Comparative analysis showing inclusion of self-education and retraining contributions in the aggregate coverage estimate (the 43.9% figure explicitly includes these channels); descriptive counts/estimates of non-degree trained entrants.
A subset of universities performs markedly better on employment effectiveness, graduate wages, and placement into popular AI roles (i.e., identifiable high-performing institutions).
Comparative analysis across the 191 universities, including employment rates, observed wage outcomes, and placement distributions; identification and reporting of key/high-performing institutions and their metrics.
Russian universities that run AI-related educational programs are contributing substantially to the national AI workforce supply.
Institutional-level monitoring data from n = 191 universities showing program enrollments, graduate counts and graduate employment into AI-related roles (descriptive analysis of supply from degree programs).
AI complements high-skill labor and raises returns to advanced cognitive and creative skills.
Microdata wage analyses and task-complementarity mappings that link AI-exposed tasks with skill groups, supported by panel regressions showing higher wages/earnings growth for higher-skill workers and by theoretical task-based models predicting complementarity.
Macroeconomic effects remain hard to observe because of a 'productivity J-curve': firms often must invest in organizational changes first and only later realize measurable financial/productivity gains from AI.
Conceptual synthesis supported by firm-level case studies and empirical papers in the reviewed literature indicating implementation lags; the brief frames this as an interpretation of mixed short-run macro evidence rather than a single causal estimate.
Organisational rules, regulatory constraints, and transparency requirements materially shape micro-level human–AI interactions and can alter adoption incentives and accountability outcomes.
Conceptual governance argument linking institutional constraints to human–AI design choices; theoretical reasoning, no empirical policy evaluation provided.
Potential productivity gains from automating routine informational tasks are conditional: net gains depend on managerial capacity to integrate AI outputs into systemic decision-making and on governance structures.
Conceptual conditional claim derived from integration of systems thinking and algorithmic optimisation literatures; no empirical measurement of productivity effects.
Information-processing and optimisation tasks exhibit clear substitution pressure (are most automatable), whereas relational and normative tasks remain complementary to human labour.
Theory-driven claim combining managerial role analysis with general automation/complementarity logic from AI economics; conceptual prediction without empirical quantification.
Human–algorithm architectures can take three forms—augment (assist), displace (replace), or reconfigure (redistribute) cognitive tasks—and their design depends on organisational design, regulation, and decision-structure rules.
Taxonomic conceptualization derived from cross-framework analysis; prescriptive mapping rather than empirical classification; no sample.
Interpersonal coordination roles (disturbance handler, liaison, leader) retain strong human elements (influence, ethics, legitimacy) that are difficult to fully algorithmise.
Conceptual argument based on the nature of relational and legitimacy-based tasks within Mintzberg’s framework and limits of algorithmic substitution; theoretical only.
Entrepreneurial and disturbance-handling roles become hybrid decision zones requiring integrated strategic and computational reasoning (modelling, simulation, anomaly detection plus contextual interpretation and values-based trade-offs).
Analytical synthesis of role demands and computational affordances; cross-framework analysis producing a hybrid strategic–computational characterization; no primary data.
Roles that rely on relational intelligence, ethical judgement, and influence (leader, liaison, figurehead, negotiator) remain primarily strategic but are increasingly supported by predictive and diagnostic analytics.
Role-specific effects derived from cross-framework conceptual mapping (Mintzberg roles × computational thinking); theoretical argumentation rather than empirical measurement.
AI systematically reconfigures managerial work by augmenting, displacing, or reconfiguring cognitive tasks across Mintzberg’s ten managerial roles.
Conceptual synthesis and comparative role mapping integrating Mintzberg’s ten managerial roles with Senge’s Five Disciplines and computational thinking; theoretical analysis only (no primary empirical data; no sample).
Commercial platforms' incentives may not align with public-interest verification, so economic policies (transparency mandates, data portability, competition policy) can reshape incentives and improve information ecosystems.
Policy implication drawn from the study's analysis of platform governance and incentive misalignment, supported by interviews and documents discussing platform interactions.
Platforms selectively adopt automated tools for triage, detection, and monitoring while keeping human judgment central to verification.
Interviews and workflow analyses indicating selective automation (for triage/monitoring) combined with human-led verification steps.
Each platform (Akeed, Teyit, Factnameh) adapts its scope and tactics according to national constraints.
Platform-level descriptions derived from interviews with staff/editors and analysis of platform outputs and workflows for each of the three organizations.
Fact-checking platforms in Jordan (Akeed), Turkey (Teyit), and Iran (Factnameh) face similar operational constraints—censorship, limited access to data, and difficulties engaging audiences—but respond with different strategies shaped by local politics.
Comparative interpretive analysis based on document analysis of platform outputs/guidelines and semi-structured interviews with staff, editors, and stakeholders from the three platforms (Akeed, Teyit, Factnameh).
The paper proposes an 'algorithmic workplace' framework emphasising hybrid agency (agents composed of humans plus GenAI), decentralised decision processes, and erosion of rigid managerial boundaries.
Conceptual synthesis derived from thematic mapping, co‑word analysis and interpretive discussion of the mapped literature; framework presented as the article's conceptual contribution.
AI diffusion and China’s delayed retirement policy jointly shape pre-retirement workers’ willingness to stay employed.
Cross-sectional survey (n=889) of pre-retirement respondents in Beijing, Guangzhou, and Lanzhou; multivariate regression analysis examining associations between employment willingness and regional AI exposure plus policy context (delayed retirement).
Passive AI use produced an initial increase in enjoyment/satisfaction that reversed once participants returned to manual work.
Pre-registered experiment (N = 269) measured enjoyment/satisfaction before and after return to manual work; passive-copy condition showed short-term increases in enjoyment/satisfaction that declined after returning to manual tasks.
AI feedback may either augment teacher productivity (complementarity) or substitute for routine teacher feedback tasks (substitution), with unclear net labor impacts.
Workshop deliberations among 50 scholars highlighting competing theoretical scenarios; no causal labor-market evidence provided.