Evidence (4892 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
Filtered →
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 |
Org Design
Remove filter
The paper derives formal conditions under which the inversion (smaller, orchestrated models outperforming frontier models) holds.
Mathematical derivations and stated sufficient/necessary conditions presented in the paper.
We develop the Institutional Fitness Manifold, a mathematical framework that evaluates AI systems along four dimensions: capability, institutional trust, affordability, and sovereign compliance.
Theoretical/model development presented in the paper (formal definition of the manifold and its four dimensions).
There have been five eras of AI development since 1943, and within the current Generative AI Era there are four distinct epochs, each initiated by a discontinuous event.
Descriptive/historical classification within the paper (counts of eras and epochs; named initiating events such as the transformer and the 'DeepSeek Moment').
Despite fears of mass unemployment, aggregate labor-market data through 2025 show limited labor-market disruption from generative AI.
Review of aggregate employment and labor-market studies and macro-level data through 2025 cited in the brief; methods include analyses of employment statistics and macro labor indicators (no single sample size reported).
We scored rule-breaking and abuse outcomes with an independent rubric-based judge across 28,112 transcript segments from multi-agent governance simulations.
Reported methodology: multi-agent governance simulations with agents in formal governmental roles, outcomes evaluated by an independent rubric-based judge; explicit sample count of 28,112 transcript segments.
Controlled experiments were run with N = 250 across five content types to validate the mechanisms.
Experimental methods reported in the paper: controlled experiments with specified sample size and content-type breakdown.
Research agenda: empirical microdata on managerial time use, task-level automation, performance outcomes, and wage impacts are needed to quantify substitution versus complementarity and to evaluate human-in-the-loop designs' effects on firm performance and distributional outcomes.
Explicit methodological recommendation within the paper; identifies gaps due to the paper's conceptual (non-empirical) approach.
There is a need for longitudinal and cross‑country empirical research to measure how hybrid work and AI tools affect promotion rates, network centrality, productivity, privacy harms, trust, and long‑term career trajectories.
Statement of research gaps derived from the paper's methodological approach (conceptual synthesis and secondary case studies) and absence of longitudinal/cross‑cultural primary data.
Practical recommendations for firms and policymakers include investing in training for AI curation/evaluation/coordination, experimenting with decentralised decision rights and governance safeguards, and monitoring competitive dynamics related to model/platform providers.
Policy and practitioner takeaways explicitly presented in the discussion/implications sections, deriving from the conceptual framework and mapped literature.
The paper recommends a research agenda for AI economists: causal microeconometric studies (DiD, IVs, RCTs), structural models with hybrid human–AI agents, measurement work on GenAI use, distributional analysis and policy evaluation.
Explicit recommendations listed in the implications and research agenda sections; logical follow‑on from bibliometric findings about gaps in causal and measurement evidence.
Bibliometric mapping profiles the intellectual structure and evolution of the field but does not establish causal effects of GenAI on organisational outcomes.
Methodological limitation explicitly stated in the paper; bibliometric approach (co‑word, citation, thematic mapping) is descriptive and historical in scope.
Co‑word and thematic analyses reveal six coherent conceptual clusters that bridge technical AI topics (e.g., LLMs, GANs) with managerial themes (e.g., autonomy, coordination, decision‑making).
Thematic mapping and co‑word network analysis performed on the 212‑paper corpus; identification of six clusters reported in results.
Bibliometric and conceptual tools (VOSviewer, Bibliometrix) were used to identify performance trends, co‑word structures, thematic maps, and conceptual evolution in the GenAI–organisation literature.
Methods section: use of VOSviewer for network visualization and Bibliometrix for bibliometric statistics, co‑word analysis, thematic mapping and Sankey thematic evolution.
The study analysed a corpus of 212 Scopus‑indexed publications covering 2018–2025 to map emergent literature on Generative AI and organisational change.
Bibliometric dataset constructed from Scopus; sample size = 212 peer‑reviewed articles; time window 2018–2025; analyses performed with Bibliometrix and VOSviewer.
Because the study is cross-sectional and self-report, causal claims are limited and generalizability is restricted to Generation Z (limitation noted in the paper).
Authors' limitations: cross-sectional/self-report design and sample restricted to Generation Z; these constraints are reported in the paper.
Study design: cross-sectional self-report survey of 450 Generation Z consumers analyzed with Structural Equation Modeling (SPSS AMOS).
Methods section reporting sample size (n = 450), target population (Generation Z), cross-sectional survey design, and analysis technique (SEM using SPSS AMOS).
The measurement and structural model show good to excellent fit and reliable constructs (CFI = 0.980, TLI = 0.974, RMSEA = 0.062, SRMR = 0.031).
Reported psychometric/model-fit indices from SEM analysis (SPSS AMOS) on sample of 450 respondents.
Outcomes reported are primarily self-reported psychological measures rather than objective productivity metrics.
Paper reports measurement instruments focused on self-reported self-efficacy, psychological ownership, meaningfulness, and enjoyment/satisfaction; no primary objective productivity metrics reported.
The experiment was pre-registered, used occupation-specific writing tasks, and employed a between-subjects design with three conditions (No-AI, Passive AI, Active collaboration).
Study design reported in the paper: pre-registration statement, N = 269, between-subjects assignment to three conditions using occupation-specific writing tasks.
Active, collaborative AI use preserves perceived meaningfulness of work at levels comparable to independent work and does not produce the lasting psychological costs seen with passive use.
Pre-registered experiment (N = 269) with post-manipulation and post-return measures; Active-collaboration condition matched No-AI on meaningfulness and showed no persistent declines after returning to manual tasks.
Active, collaborative AI use preserves psychological ownership of outputs at levels comparable to independent work.
Pre-registered experiment (N = 269); Active-collaboration condition reported ownership levels similar to No-AI condition on self-report scales.
Active, collaborative AI use (human drafts first, then uses AI to refine) preserves self-efficacy at levels comparable to independent (no-AI) work.
Pre-registered experiment (N = 269) comparing Active-collaboration and No-AI conditions; no statistically meaningful differences in self-efficacy between them (self-reported measures).
The authors propose research priorities for economists: quantify productivity gains from closing the actionability gap; estimate firm-level heterogeneity in evaluation capability and its effect on adoption; and model investment trade-offs between building evaluation-to-action pipelines versus accepting reduced LLM performance.
Paper's concluding recommendations for future research directions (explicitly listed by the authors).
The paper produces as primary outcomes a taxonomy of ten evaluation practices, the articulation of the results-actionability gap, and recommended strategies observed among successful teams.
Authors report these as the main outcomes of their thematic analysis and syntheses from the 19 interviews.
The study method consisted of semi-structured qualitative interviews with 19 practitioners across multiple industries and roles, analyzed via thematic coding.
Explicit methods section of the paper stating sample size (n=19), participant diversity, interview approach, and coding/analysis procedure.
The analysis used sentence‑transformer models to produce dense vector representations of article text and UMAP to project those embeddings into a low‑dimensional thematic map for cluster identification and gap detection.
Methods section specifying use of sentence‑transformer embeddings and UMAP for dimensionality reduction/visualization of article text.
The study followed a PRISMA protocol for literature selection and included peer‑reviewed journal articles published between 2014 and 2024, with a final sample size of n = 109.
Explicit methodological statement in the paper describing the literature search, inclusion/exclusion criteria, and final sample.
Twenty‑seven papers study marketing in banking without using NLP methods.
PRISMA systematic review; categorization of the 109 selected articles into the three coverage groups (8, 74, 27).
Seventy‑four papers study NLP in marketing more broadly (not specifically banking).
Same PRISMA‑based systematic review and manual categorization of the final sample n = 109 into topical buckets (NLP in marketing vs. NLP in bank marketing vs. marketing in banking without NLP).
Only 8 peer‑reviewed papers directly examine NLP in bank marketing (out of a final sample of 109 articles published 2014–2024).
Systematic review following PRISMA protocol; final sample n = 109 peer‑reviewed journal articles published 2014–2024; manual screening and categorization yielding counts by topic.
The study's findings are qualitative and case-driven (Xiaomi and Deloitte); generalizability is limited by case selection and the absence of standardized quantitative metrics.
Methods section explicitly states case analysis and literature review as primary methods and notes lack of large-scale quantitative measurement.
The study is qualitative and law-focused and uses Vietnam as a focused case study without collecting primary quantitative field data.
Explicit Data & Methods statement in the paper indicating doctrinal legal analysis, comparative institutional analysis, and normative framework development; no primary quantitative sample.
The study recommends empirical metrics for future evaluation of reforms, including processing time per case, reversal rates on appeal, administrative litigation frequency, compliance and procurement costs, investment flows into public-sector AI, and changes in labor composition and wages in administrative agencies.
Methodological recommendation arising from the paper's normative and comparative analysis.
The paper's argument is principally theoretical and prescriptive and requires empirical validation across domains and at scale.
Author-stated limitation in the Data & Methods section noting that the work is primarily conceptual and that empirical validation is needed.
Operationalizing DSS requires building domain ontologies/knowledge graphs, designing synthetic curricula, training compact domain models, benchmarking against monolithic LLMs, and measuring total cost-of-ownership (energy, latency, bandwidth, infrastructure).
Paper's recommended experimental and measurement agenda (procedural/methodological prescriptions); this is a proposed research plan rather than an empirical result.
The paper does not claim proprietary deployment metrics beyond qualitative field observations; experimental formalizations are provided for reproducible evaluation instead.
Authors explicitly note they document how to reproduce experiments but do not claim proprietary deployment metrics beyond qualitative field observations.
The paper recommends tracking specific operational and economic metrics: MTTR for tool failures, per-invocation latency variance, per-interaction operational cost, frequency of identity-related incidents, human remediation hours per 1,000 incidents, and SLA breach rates.
Explicit list of recommended metrics in the implications and metrics-to-track sections of the paper.
The paper provides a production-readiness checklist and instructions for reproducible evaluation alongside the proposed mechanisms.
Deliverables enumerated in the paper include a production-readiness checklist and reproducible experimental methodology.
All three proposed mechanisms (CABP, ATBA, SERF) are formalized as testable hypotheses with reproducible experimental methodology (benchmarks, latency/error models, broker pipeline semantics).
Paper includes formal descriptions and reproducible evaluation instructions and benchmarks; authors state methods to reproduce experiments are provided.
The paper organizes production failure modes across five dimensions—server contracts, user context, timeouts, errors, and observability—and provides concrete failure vignettes from an enterprise deployment.
Taxonomy and failure vignettes are listed as design artifacts and deliverables in the paper; derived from observational analysis of production logs and incidents.
Sample sizes reported: human–AI experiment n = 126; human–human benchmark n = 108.
Study's Data & Methods section reporting sample sizes for the human–AI experiment (n = 126) and citing the human–human benchmark (Dvorak & Fehrler 2024, n = 108).
Experimental design: subjects played an indefinitely repeated Prisoner’s Dilemma in supergames with two between-subjects treatments varying chat timing (chat only before first round of each supergame vs chat before every round); the AI partner was GPT-5.2.
Methods description of the lab experiment reported in the paper: indefinitely repeated PD in supergames, two chat-frequency between-subjects treatments, AI implemented as GPT-5.2; human–AI sample n = 126.
Allowing repeated pre-play communication (chat before every round) has no detectable effect on cooperation rates when the partner is an AI.
Between-subjects manipulation within the human–AI experiment comparing chat-before-first-round vs chat-before-every-round treatments (human–AI n = 126 total); statistical comparison of cooperation rates across the two chat-frequency treatments showed no detectable difference.
Initial cooperation rates against the AI (GPT-5.2) are high and comparable to initial cooperation in human–human pairs.
Laboratory experiment with human subjects playing an indefinitely repeated Prisoner’s Dilemma against an AI chatbot (GPT-5.2); human–AI sample n = 126; human–human benchmark taken from Dvorak & Fehrler (2024) with n = 108; comparison of initial-round / early-round cooperation rates across conditions.
Suggested empirical research directions for AI economists include: comparing LLM performance and economic outcomes on rule‑encodable vs tacit tasks; quantifying performance decline when forcing LLMs into interpretable rule representations; studying contracting/pricing where buyers cannot verify internal rules; and measuring returns to scale attributable to tacit capabilities.
Explicitly enumerated recommended research agenda items in the paper; these are proposed studies rather than executed work.
New metrics are needed to value tacit capabilities — e.g., measures of transfer, generalization under distribution shifts, ease of integrating with human workflows, and irreducibility to compressed rule representations.
Methodological recommendation in the paper listing specific metric categories for future empirical work.
Suggested empirical validations (not performed) include benchmarking LLMs versus rule systems on allegedly rule‑encodable tasks, attempting rule extraction and measuring fidelity loss, and compression/distillation studies to quantify irreducible task performance.
Recommendations and proposed experimental directions listed in the paper; these are proposals, not executed studies.
The paper contains mostly qualitative and historically grounded empirical content and reports no primary datasets or large‑scale experimental results in support of the formal thesis.
Explicit declaration in the Data & Methods section that empirical content is qualitative/historical and no new datasets were collected.
The paper's core methodological approach is conceptual and theoretical argumentation (formal/logical proof, historical examples, and philosophical framing), not empirical experimentation.
Stated Data & Methods description indicating reliance on formal logic, historical case analysis, and philosophical argument; absence of primary datasets.
Measuring the marginal cost of runtime governance, the tradeoff curve between task completion and compliance risk, and calibrating violation probabilities are open empirical research questions identified by the paper.
Explicit list of open problems and proposed empirical research agenda in the Implications/Measurement sections of the paper.