Evidence (4004 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
Filter claims →
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
Filtered →
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 |
Labor Markets
Remove filter
Automated compliance and credentialing systems raise governance issues (auditability, appeals mechanisms) and risk incorrect automated deregistration if not properly governed.
Governance and algorithmic-risk discussion in the paper; logical argumentation rather than case-based evidence.
The paper models career progression as a continuous function and treats certification gaps as discontinuities that impede labour-market mobility.
Mathematical/conceptual modeling described in the methods (career-progression-as-continuous-function approach); this is a modeling choice reported in the paper rather than an empirical finding.
There is limited long-term impact evidence and few system-level assessments of AI in developing-country agriculture.
Authors' methodological caveat based on the temporal scope and types of studies available in the >60-study review.
Substantial compute and resource requirements for training and inference concentrate capabilities among well‑resourced labs and firms.
Paper discusses large compute budgets for training/inference and states that performance scales with data, model size, and compute; it infers concentration of capabilities but provides no empirical market concentration measures.
Structure predictors depend on training data and exhibit biases; experimental validation remains necessary.
Paper notes dependence on training data biases and the need for experimental validation; references data sources (PDB, UniRef, metagenomic catalogs) but does not quantify bias magnitudes.
Current limitations include inaccurate prediction of multi‑chain complexes, flexible or rare conformational states, and limited prediction of dynamic ensembles.
Paper explicitly enumerates these limitations in the 'Ongoing limitations' section; no quantitative failure rates are given.
Traditional computational methods struggle without homologous templates or with complex folding/dynamics.
Paper discusses limitations of traditional computational methods, emphasizing dependence on homologous templates and difficulty with complex folding/dynamics; specific method comparisons or sample sizes are not provided.
Opacity, bias, and errors in AI systems demand auditing, standards, and governance (algorithmic accountability) to ensure trustworthy assessment.
Synthesis of literature on algorithmic bias and accountability plus policy analysis recommending audits and standards; supported by country cases that discuss governance concerns.
Student data used by AI vendors raises risks around consent, reuse, commercial exploitation, and other data-privacy concerns.
Policy analysis and literature on data governance, privacy law debates; examples from national policy documents in the comparative cases. No original data on breaches or misuse presented.
Empirical evaluation of integrated defenses, quantitative cost/benefit analyses, and standardized threat models for VR are research gaps that remain unaddressed in the literature window surveyed (2023–2025).
Authors' stated limitations from their comparative literature review of 31 studies noting an absence of primary empirical validation and quantitative economic analyses in the reviewed corpus.
Immersive VR systems collect continuous multimodal signals (motion tracking, gaze, voice, biometrics) that enable novel inference, spoofing, and manipulation attacks beyond traditional IT threats.
Synthesis of threat descriptions across the 31 reviewed peer‑reviewed studies (2023–2025) documenting sensor modalities and attack vectors; qualitative comparative evaluation of attack surfaces.
Limitations of the study include reliance on self-reported perceptions (subject to response and survivorship bias), lack of experimental/causal identification, potential non-representative sample, and cross-sectional design limiting inference about long-term productivity effects.
Authors' stated limitations in the paper summary.
Tasks that are routine, repetitive, or pattern‑based (e.g., boilerplate coding, refactoring, unit test generation, some accessibility fixes) will be increasingly automated by AI.
Task‑level decomposition and examples of current automation capabilities (code generation, test suggestion tools); conceptual projection rather than empirical measurement.
Common barriers to effective RM implementation include siloed functions/weak coordination, limited resources or expertise, poor data quality/lack of metrics, and cultural resistance driven by short-term incentives.
Frequent identification of these barriers across the reviewed literature and practitioner sources synthesized via thematic analysis over the last ten years.
Hierarchy compresses: fewer organizational layers are needed for a given firm output as coordination costs fall.
Analytical proposition in the theoretical model and simulation results showing reduced number of layers under coordination compression.
A one standard-deviation increase in AI adoption (2019–2025, 38 OECD countries) causally reduces employment in routine cognitive occupations by 2.3%.
Panel of 38 OECD countries, 2019–2025; AI Adoption Index (composite of enterprise AI investment, AI patent filings, workforce/firm AI-use surveys); instrumental-variable (IV) estimation to identify causal effect on occupational employment; country and year fixed effects and macro controls reported.
Higher measured GDP need not imply higher aggregate welfare: the private costs of the arms race can outweigh the market gains from increased output.
Welfare comparisons performed in the model showing parameter regions where private equilibrium raises GDP but reduces aggregate welfare once investment costs are included.
Because private incentives push agents toward tail outcomes, aggregate overinvestment occurs relative to the social optimum (the arms race is inefficient).
Welfare calculations and comparison of private vs social optima within the model; the paper shows private equilibrium investment exceeds the socially optimal investment given the externalities of the arms race.
Upfront costs for AI adoption are substantial: development, clinical validation, regulatory compliance, EHR integration, and ongoing monitoring.
Implementation and regulatory literature synthesized in the review documenting typical cost categories and reported expenditures for clinical AI projects.
Large language models (LLMs) suffer from hallucinations (fabricated facts), overconfidence, and unpredictable failure modes in open-ended tasks.
Technical papers and benchmarks on LLM factuality, calibration, and failure modes summarized in the review; empirical evaluations showing instances of fabricated outputs and calibration issues.
Contemporary AI systems have no capacity for physical examination, sensorimotor procedures, or direct patient-contact diagnostics.
Technical limitations of CNNs and LLMs described in literature (lack of embodiment, no sensorimotor capabilities) and absence of credible empirical demonstrations of safe autonomous physical clinical procedures in reviewed studies.
Current models exhibit poor out-of-distribution (OOD) generalization: performance degrades when inputs differ from training distributions.
Technical literature and robustness/domain-shift research reviewed in the paper documenting declines in model accuracy under domain shift and dataset changes.
Risks include bias and discrimination, opacity in decision-making, privacy and cybersecurity threats, liability gaps, and uneven distribution of benefits that can exacerbate inequality.
Compilation from academic and policy literature, regulatory gap analyses, and examples of problematic AI use cases identified in the report's sectoral review.
AI creates significant ethical, legal and distributional risks.
Review of policy documents, academic and policy literature, and documented examples of AI deployment across multiple sectors highlighting harms (bias, privacy breaches, liability gaps, unequal benefits).
Reliance on imperfect data and model assumptions can produce biased or misleading forecasts; careful validation, transparency about assumptions, and governance are necessary.
Risks & governance discussion in the paper raising this limitation and recommending practices (qualitative argumentation).
AI-generated code can introduce security vulnerabilities and raise licensing/intellectual-property concerns.
Case studies of security incidents, analyses of generated code provenance, and vulnerability-detection studies synthesized in the review.
LLMs sometimes generate incorrect, nonsensical, or insecure code (hallucinations).
Multiple benchmarks, code-generation accuracy tests, and incident case studies documented in the empirical literature showing incorrect or fabricated outputs.
Data security, privacy risks, unequal gains, and regulatory shortfalls can undermine the benefits of AI/robotics adoption.
Policy and risk analyses from secondary literature, case studies, and institutional reports synthesized in the paper; examples cited but no original incident-level dataset or incidence rates provided.
Transition frictions and skills mismatches are important barriers to workers moving into newly created AI‑related roles.
Qualitative review of workforce and skills literature, case studies, and sector reports; evidence comes from secondary sources with varied methodologies; the paper does not report pooled quantitative estimates.
International and national legal approaches to these stages are fragmented, creating uncertainty for IP, privacy, liability and evidence law.
Comparative review of international and national legal approaches and judicial responses cited in the paper (secondary legal sources).
Output-stage risks include authenticity/deception concerns, attribution and reuse-rights disputes, reputational harms, and broader societal impacts from abundant generated media.
Review of empirical studies on media authenticity, legal cases, and policy analyses included in the narrative review.
Process-stage risks include governance of model development, control over deployment, transparency, auditing, and operational safety.
Conceptual synthesis of technical governance literature and policy reports cited in the narrative review.
Input-stage risks include concerns about consent, copyright, representativeness, bias, provenance and data ownership for training material.
Synthesis of legal and policy literature and documented legal cases/statutes related to training data and IP/privacy issues (secondary sources only).
Generative audiovisual AI poses material ethical, control, transparency and legal challenges across three stages — input (training data), process (development & deployment), and output (use of artifacts).
Conceptual three-stage framework built from comparative review of literature, legal cases/statutes and policy reports described in the paper.
Limitations of the study include potential selection bias in reviewed sources and contingency of conclusions on evolving legal decisions and technology developments.
Author-stated limitations section within the paper; qualitative acknowledgement rather than empirical bias assessment.
Output-stage risks include challenges to authenticity and provenance, erosion of trust (deepfakes and misinformation), and potential legal liability for harms caused by generated content.
Synthesis of technical papers on deepfakes, legal analyses of liability, and policy reports referenced in the review; no original incident dataset or quantitative prevalence estimate included.
Input-stage risks include copyright infringement, lack of consent, poor data provenance, and biases/representational harms encoded in training datasets.
Review and synthesis of academic and legal literature on training data issues; examples and case law discussed, but no original dataset audit or sample counts provided.
Use of these models faces significant ethical, control, transparency, and legal challenges across three stages—input (training data), process (development/control), and output (generated artifacts).
Framework constructed from interdisciplinary literature (technical, ethical, legal sources) and review of statutes/judicial approaches; qualitative synthesis rather than primary data.
High environmental constraints in many African regions (poor infrastructure, challenging geography, frequent climate shocks) materially affect logistics, resilience, and supply-chain performance.
Review of literature on infrastructure, geography, and climate impacts in the conceptual paper.
Africa is abundant in natural resources but exhibits relatively low development/outcomes from those resources, creating resource allocation and value-capture problems relevant to OSCM.
Development economics and regional studies literature cited in the paper's synthesis; conceptual claim without new empirical testing.
Africa has a large informal economy and many informal organizations that shape supply-chain behavior and market functioning.
Literature synthesis citing development and institutional studies (no primary data collection in the paper).
Results reflect small-scale e-commerce use cases; external validity to larger firms, other sectors, or more complex tasks is not established.
Scope of deployments limited to small-scale e-commerce settings as stated in methods; no cross-sector or large-firm samples reported in summary.
The study's evidence is observational rather than randomized controlled trials, so causal estimates about productivity impacts are suggestive rather than definitive.
Declared study design: applied experimentation and observational analysis of deployments (no randomized assignment); methods section explicitly notes observational limitation.
Deficits in governance, auditing, and interpretability constrain the safe deployment of generative AI in firms.
Synthesis of industry reports and conceptual literature noting gaps in governance and interpretability; no quantitative governance dataset reported.
Algorithmic biases in generative AI can amplify and codify discriminatory patterns in organizational decisions.
Extensive literature on algorithmic bias synthesized in the review and applied to generative models; case examples referenced.
Generative AI use introduces significant organizational risks including data privacy breaches and leakage when models or third‑party services are used.
Conceptual analysis and references to documented incidents and industry reports within the review; no single aggregated incident dataset provided.
Generated code can introduce security vulnerabilities.
Security analyses and code audits documenting examples where LLM-generated code contains known vulnerability patterns; incident-oriented case studies and controlled experiments assessing vulnerability incidence.
LLMs can produce plausible-looking but incorrect or insecure code (so-called 'hallucinations').
Benchmarks and controlled tests demonstrating incorrect outputs; security analyses and replicated examples showing erroneous or insecure snippets produced by LLMs across multiple models and prompts.
The technical feasibility of robust token verification and resistance to spoofing needs demonstration; it is not yet proven.
Authors explicitly acknowledge this limitation in the paper; no prototypes or red-team results are presented.
AI-driven impacts will be heterogeneous across education, race, gender, age, firm size, and geography, implying crucial equity concerns and the need for disaggregated reporting and targeted validation.
Policy analysis and literature synthesis in the paper; this claim reflects widely-documented labor economics findings about heterogeneous technological impacts though no new empirical breakdowns provided here.