Evidence (3714 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
20058 claims
Filter claims →
Productivity
17184 claims
Filter claims →
Governance
16099 claims
Filter claims →
Human-AI Collaboration
16034 claims
Filter claims →
Innovation
10501 claims
Filter claims →
Org Design
10496 claims
Filter claims →
Labor Markets
6444 claims
Filter claims →
Skills & Training
5385 claims
Filter claims →
Inequality
4148 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 | 1820 | 479 | 278 | 1820 | 4588 |
| Organizational Efficiency | 2711 | 616 | 401 | 173 | 3922 |
| Governance & Regulation | 2075 | 886 | 459 | 246 | 3714 |
| Technology Adoption Rate | 1467 | 530 | 258 | 206 | 2488 |
| Decision Quality | 1281 | 496 | 289 | 152 | 2228 |
| Output Quality | 1227 | 447 | 207 | 138 | 2025 |
| AI Safety & Ethics | 634 | 754 | 207 | 83 | 1688 |
| Research Productivity | 826 | 241 | 114 | 422 | 1624 |
| Firm Productivity | 1052 | 154 | 163 | 66 | 1441 |
| Task Allocation | 685 | 211 | 331 | 99 | 1335 |
| Market Structure | 433 | 423 | 242 | 46 | 1150 |
| Innovation Output | 639 | 91 | 105 | 34 | 871 |
| Task Completion Time | 476 | 113 | 43 | 36 | 672 |
| Firm Revenue | 445 | 126 | 58 | 25 | 656 |
| Skill Acquisition | 364 | 119 | 109 | 34 | 626 |
| Consumer Welfare | 288 | 167 | 104 | 31 | 592 |
| Employment Level | 214 | 140 | 174 | 50 | 582 |
| Error Rate | 230 | 251 | 35 | 16 | 535 |
| Fiscal & Macroeconomic | 268 | 136 | 71 | 50 | 532 |
| Inequality Measures | 100 | 307 | 96 | 12 | 515 |
| Worker Satisfaction | 221 | 173 | 60 | 30 | 484 |
| Automation Exposure | 155 | 138 | 65 | 36 | 398 |
| Regulatory Compliance | 171 | 120 | 30 | 13 | 335 |
| Developer Productivity | 222 | 58 | 27 | 13 | 321 |
| Team Performance | 188 | 56 | 50 | 24 | 320 |
| Wages & Compensation | 146 | 104 | 46 | 16 | 312 |
| Training Effectiveness | 207 | 41 | 21 | 26 | 298 |
| Job Displacement | 23 | 153 | 52 | 4 | 232 |
| Hiring & Recruitment | 102 | 57 | 30 | 11 | 202 |
| Skill Obsolescence | 16 | 102 | 24 | 6 | 148 |
| Creative Output | 71 | 42 | 23 | 6 | 143 |
| Social Protection | 57 | 30 | 11 | 3 | 101 |
| Labor Share of Income | 29 | 42 | 24 | 2 | 97 |
| Worker Turnover | 43 | 29 | 6 | 4 | 82 |
| Industry | — | — | — | 1 | 1 |
Platform regulations such as transparency requirements, auditability, and algorithmic impact assessments can change the incentives and equilibrium outcomes associated with algorithmic design in news markets.
Theoretical policy implication; the article proposes evaluating such effects but does not estimate them.
Accountability for public knowledge is distributed across journalists, platforms, and regulators rather than being assigned to a single actor.
Theoretical account of accountability within the contested epistemic space; no observed cases or measured accountability outcomes are reported.
The intersection of editorial, algorithmic, and regulatory logics forms a contested epistemic space in which visibility, credibility, and accountability are negotiated and may be stabilized or destabilized.
Theoretical framework identifying the intersection and its effects on public knowledge; no empirical validation is presented.
Journalism's epistemic authority emerges from the interaction of editorial, algorithmic, and regulatory logics rather than being produced solely within newsrooms.
Conceptual synthesis in the article's main finding; no empirical sample or causal test is reported.
The study advocates balancing automation gains with ethical safeguards and human oversight in AI-driven talent acquisition.
Qualitative exploration of AI-enabled recruitment experiences, including participants' concerns about privacy, organisational readiness, and ethical governance.
The interaction of the AI Act and GDPR creates synergies in governance and accountability but can also compound compliance burdens, producing trade-offs between reduced harms and trust on one hand and efficiency and innovation on the other.
Synthesis of the doctrinal analysis and the paper's discussion of compliance costs, trust, innovation, and regulatory interaction; no quantitative trade-off estimate is provided.
EU rules may diffuse globally as de facto standards because firms adopt uniform EU-aligned practices to preserve market access, potentially reducing regulatory fragmentation while exporting EU norms.
Comparative legal analysis of extraterritoriality and market-access incentives, interpreted through the regulatory-governance and regulatory-capitalism literature.
Organizational AI acculturation can generate positive private returns while also producing negative social externalities through propagation of biased or stale practices.
Conceptual analysis of the divergence between firm-level incentives and aggregate welfare; the paper proposes studying incentive and regulatory designs rather than reporting an empirical estimate.
The management responses associate AI-related employment change with distributional consequences, institutional governance and organisational responsibility, not only productivity.
Coding of 159 usable Q23 management responses identified social fairness, industry regulation, data privacy, human–AI collaboration, skills training and employee welfare as recurring categories.
The company-level version of SCAF conceptualizes vulnerability as exposure created by product deployment reach, coping as features that help users respond to and recover from failures, and adaptive capacity as features and practices that reduce future failures and improve learning.
Table 2's operational definitions of company contributions to vulnerability, coping, and adaptive capacities.
Four of the five primitives were built and running in private pilots, while supply-chain functionality was built as separate tooling but had not yet been integrated into the request path.
Authors' implementation-status report; no pilot sample size or outcome measurements are reported.
The paper limits PES's applicability to settings in which multi-user deployment, execution audit, and expected persona churn occur jointly.
Scope and applicability statement; this is a stated boundary condition rather than an empirically validated result.
The adoption drivers identified in prior frameworks should not be treated as parallel, co-equal factors in mandate-driven public-sector organizations; instead, their effects are hierarchically conditioned by political authorization.
The study's comparative qualitative findings across leading and emergent governments, interpreted through Task–Technology Fit, Diffusion of Innovations, and implementation theory.
When an AI algorithm makes decisions affecting more than one person, alignment is a social-choice problem because the system must reconcile and aggregate divergent individual preferences.
Conceptual framework developed in the abstract and introduction; the paper formalizes alignment as preference aggregation over welfare impacts.
The feasibility and welfare effects of share compensation depend on firm size, share liquidity and valuation, bargaining and negotiation mechanisms, corporate governance, legal and regulatory constraints, and transaction costs.
The paper acknowledges implementation limitations and institutional frictions rather than testing them empirically.
Regulatory and institutional responses, including intellectual-property law, data-protection regimes, and antitrust enforcement, interact with entrepreneurial actions to shape whether AI property regimes are private, public, or hybrid.
Conceptual institutional analysis and proposed AI-governance research agenda.
Corporate narratives portray digital twins and industrial AI as enabling synchronized, centrally managed, and predictive factory operations, but observed shop-floor practice departs from this framing.
Discourse analysis of promotional texts, vendor claims, and managerial presentations contrasted with ethnographic observation and worker interviews.
The paper proposes six theory-grounded propositions and a five-stage implementation pathway, while calling for auditable, longitudinal, multi-country empirical testing.
Reported output and research agenda of the transparent integrative literature review.
Strong data governance, consent mechanisms, and secure data-sharing architectures are necessary to obtain the benefits of AI-enabled finance while limiting cybersecurity and privacy harms.
Conceptual synthesis of governance, privacy, security, and data-sharing requirements identified as boundary conditions in the review.
The paper’s completeness guarantee is limited to mediated channels; unmediated channels remain an explicit residual assumption.
Threat-model qualification stating that the architecture detects omissions only where the witness boundary and per-channel sequencing apply.
HANSARD reports compensation-set size rather than the conventional Chockler-Halpern degree of responsibility, because its modified counterfactual definition does not require setting variables to non-actual values.
Conceptual comparison between the modified Halpern-Pearl definition and the conventional responsibility measure 1/(1+k).
Higher online engagement and more effective state controls coexist in a context where online mobilization persists but its translation into mass offline protest is dampened.
Comparative historical and qualitative analysis, supported by Mo Ibrahim Foundation indicators and synthesis of documented protest and online-repression cases.
Neither a blanket prohibition of opaque AI in auditing nor uncritical reliance on opaque AI is defensible; regulation should be calibrated to the risks created by opacity rather than to AI use as such.
Normative legal and regulatory argument drawing on comparative analysis of the EU AI Act, IAASB and PCAOB developments, and common-law liability doctrine.
The framework separates evidence into public-data, customs-internal, and gateway-conditional tiers, and uses Bangladesh as an illustration of how legal access and admissibility constraints shape feasible data flows and operators.
Institutional and statute-level legal analysis; jurisdictional transferability is explicitly limited.
The implementation of the Rider Law was shaped by interactions among platform strategies, algorithmic technologies, worker bargaining capacity, and state enforcement capacity, rather than by the statute alone.
Qualitative case-study analysis focused on enforcement processes, platform counter-strategies, worker organisation, and state capacity.
US-led restrictions targeting China change access conditions but do not substitute for corporate control over the semiconductor production pipeline.
Process tracing of US-led restrictions alongside corporate control over lithography services, qualification, packaging, and capacity scheduling.
State-leveraged chokepoints and corporate chokeholds are distinct governance mechanisms: states govern access to frontier technologies and inputs, while firms govern the operational continuity and tempo of production.
Conceptual framework and process-tracing comparison of export controls with corporate control over servicing, qualification, packaging, and capacity scheduling.
Blockchain creates new assurance targets involving consensus-protocol correctness and governance, cryptographic key management and custody, smart-contract correctness and vulnerabilities, and oracle reliability.
Standards analysis identifying new assurance objects created by blockchain systems and their links to audit evidence.
Treating frontier AI as scientific infrastructure implies that AI-access decisions shape the long-run distribution of scientific capability and power.
Conceptual interpretation of the model's endogenous feedback between AI access, credit, resources, future access, and laboratory participation.
AI diffusion may accelerate productivity while also increasing risks related to competition, market concentration, and strategic dependencies, requiring complementary governance measures.
Policy caution based on the paper's institutional and industrial-policy analysis; no direct productivity or concentration estimates are reported.
Which long-run ownership regime occurs is determined by law and initial conditions rather than by technology alone.
The paper’s theoretical ownership analysis treats legal ownership arrangements and initial ownership as determinants of ε_t and the terminal regime.
The benefits of AI integration in scientific communities are not automatic; they depend on institutional adaptation and the development of new standards and practices.
Model interpretation and simulation findings linking AI effectiveness to organizational practices, including standardization, protocols, divisions of labor, and documentation of tacit factors.
The individual, legal, non-human configuration is controversial when applied to AI because it raises the question of whether an AI system could itself be legally liable or accountable.
Conceptual analysis of the typology, drawing on legal theory concerning animals and qualified legal standing.
Optimizing clinical AI for short-term satisfaction or engagement can be economically attractive but clinically and socially suboptimal.
Conceptual analysis of incentive misalignment between engagement metrics and long-term clinical outcomes.
Hybrid human-AI decision structures are recommended for the large majority of consequential business decisions under current AI capabilities.
Organizational decision-making typology by Shrestha, Ben-Menahem, and von Krogh, summarized in the review; the recommendation is based on decision frequency, stakes, and specification of the prediction task.
The framework is intended as an additional control layer and does not replace permission gates or sandboxing.
The paper's stated system-design conclusion and limitation: learned restraint complements, rather than substitutes for, brokered permission controls and isolated execution environments.
DMM and gold-standard-label methods such as DSL and PPI are complementary: DMM avoids assumptions about gold-standard labels but requires stronger assumptions about measurement errors, whereas validation-based methods make fewer assumptions about measurement errors but require gold-standard labels.
The paper explicitly contrasts the identifying assumptions and data requirements of DMM with DSL and PPI.
Multimodal AI-centered data management creates technical and governance challenges involving scalability, interpretability, privacy, provenance, access control, compliance, and interoperability.
Conceptual discussion of system limitations and governance requirements; no quantitative evaluation of the challenges is reported.
Embedding inheritance enforcement on-chain would require trusted or semi-trusted external institutions, including death-attestation registries, proof-of-humanity systems, attesters, challenge procedures, and courts or notaries.
Speculative escalation game in which the protocol is hardened through independent attesters, challenge windows, time locks, biometric enrollment, social vouching, and state identity bridges.
Islamic and Catholic concerns converge substantially around dignity, responsibility, and social justice, but Islamic intellectual resources generate distinct emphases with implications for policy and institutional design.
The essay performs a comparative ethical framing of the encyclical and Islamic intellectual traditions rather than a systematic point-by-point comparison.
Islamic intellectual traditions raise foundational questions about personhood, knowledge, work, responsibility, and dignity in the context of large, complex systems, while framing them differently from Pope Leo XIV's Magnifica Humanitas.
The essay uses hermeneutic interpretation of the encyclical alongside textual and conceptual analysis of Islamic theological, juridical, and ethical traditions.
Government responses that suppress demand, such as fees, are generally faster to deploy than measures that increase service capacity, but they can reduce service quality and accessibility and introduce procedural inequality.
The paper's mapping of response options and review of precedent; demand-suppression examples include fees, rate limits, and in-person requirements, while capacity measures include staffing, AI deployment, and interface redesign.
Perceived legitimacy is a boundary condition for guardrail effectiveness: users are more likely to circumvent guardrails they regard as opaque or disproportionate.
The study identifies perceived legitimacy from observed workaround behavior in the quasi-experimental field deployment and presents it as a theoretical boundary condition.
Institutional capacity moderates the extent to which AI-related technological inputs translate into productive and inclusive outcomes.
The paper's integrated theoretical framework treats governance, regulation, skills, and social-protection systems as moderating conditions across productivity, distribution, and welfare channels.
Risk-based disclosure requirements for high-impact AI have ancillary value, whereas universal labeling of all GenAI outputs fails to differentiate between low-risk and high-risk uses.
Comparative analysis of Korea's AI Basic Law and its distinction between high-impact AI obligations and universal GenAI notification and labeling; no empirical assessment is reported.
The review proposes a multidimensional framework integrating behavioural, institutional, and technological perspectives to explain persistent audit failures.
Qualitative thematic synthesis and integration of recurring factors into a conceptual framework; no meta-analytic statistical pooling was reported.
The four impact factors interact in a cyclical and adaptive process that continually transforms platform capabilities, governance, and participant incentives.
Theoretical framework developed from inductive coding of the qualitative interview data.
Deploying trustworthy, scalable, and privacy-respecting IDSS requires addressing data quality, explainability, computational costs, governance, and user trust.
Qualitative review identifying cross-cutting implementation challenges across surveyed techniques and application domains.
Worker participation and co-design shape whether AI produces equitable, productivity-enhancing outcomes or harmful displacement and deskilling.
The paper presents participation and governance as the process-level moderators determining whether AI functions as a complement or substitute for workers' skills.
Subsidizing AI tools alone may have limited impact, whereas interventions that build managerial capacity, data infrastructure, workforce skills, and governance may be more effective.
Policy implication derived from the framework’s emphasis on complementary organizational capabilities; not directly tested through a policy intervention.