Evidence (9875 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
Filtered →
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
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 |
Adoption
Remove filter
Industry-wise, sectors with higher levels of digitalization (e.g., mining, finance, energy) show stronger income effects, while traditional sectors (e.g., agriculture, public services) show limited impact.
Industry-level heterogeneity analysis in the two-way fixed effects panel using provincial data (2011–2021), reporting larger estimated effects for high-digital sectors and small or null effects for traditional sectors.
Regionally, eastern provinces experience greater income gains from digital development than central and western provinces.
Regional heterogeneity results from the paper's two-way fixed effects panel (31 provinces, 2011–2021) comparing estimated effects across eastern, central, and western regions.
AI-flagged complaints are geographically unevenly distributed.
Geographic analysis of AI-flagged complaint shares across jurisdictions using case metadata; authors report uneven distribution.
As a representative of new quality productive forces, brain–computer interface (BCI) technology raises high expectations but also acute concerns about brain‑privacy protection.
Statement in paper's introduction/abstract; conceptual observation based on literature and contextual analysis (no empirical study reported).
The optimal architecture is highly task-dependent.
Empirical claim in the abstract: experiments across tasks showed that different hybrid architectures perform best for different tasks.
Task accuracy, monetary cost, and edge energy consumption are tightly coupled in hybrid MAS design.
Claim made in the abstract and investigated empirically by adapting MAS architectures and measuring power, cost, and performance trade-offs.
Any measurement of AI brand perception must condition on the buyer persona supplying the query: the same prompt produces materially different recommendation sets depending on who the model thinks is asking, and a measurement protocol that aggregates across personas systematically obscures that variation.
Argument based on observed persona-driven variation in recommendation sets across the audit; policy/methodological recommendation derived from empirical results.
The Anthropic model shows a larger point-estimate effect than the OpenAI configurations, though clustered CIs overlap for the closer contrast (sonnet vs. OpenAI/high).
Comparison of point estimates and clustered confidence intervals across model configuration cells in the audit.
No single LLM dominates across engine types, highlighting the importance of specific tasks and tradeoffs between speed and accuracy.
Empirical observation from cross-engine evaluations reported in the paper; descriptive conclusion without numeric dominance metrics or sample sizes in the excerpt.
The evaluations implemented by the initiative demonstrate that AI enabled modeling tools perform better at discussion and basic qualitative tasks than with causal reasoning and quantitative error fixing.
Result reported from the implemented evaluations comparing relative performance across task categories (discussion/qualitative vs causal reasoning/quantitative error fixing); no quantitative effect sizes or sample sizes provided in the excerpt.
When engines from the sd ai project are coupled with different LLMs, their performance on these evaluations reveals variability across different AI tools.
Empirical statement in the paper based on applying the implemented evaluations to different engine+LLM combinations; no numeric performance metrics or sample sizes reported in the excerpt.
We illustrate this transition through examples in consumer markets, education, news, and coding.
Authors state they use sectoral examples to illustrate the framework; this is a claim about the paper's contents rather than an empirical finding.
We offer a three-stage lens: Augmentation, Automation, and Reconstruction.
Conceptual framework proposed by the authors; presented as a taxonomy in the paper (no empirical validation reported in the excerpt).
Human capital structure moderates the relationship between AI application and enterprise innovation efficiency.
Moderation analysis on A-share listed firms (2012–2023) indicating significant interaction effects between AI application and measures of human capital structure.
Fiscal support intensity moderates the impact of AI application on enterprise innovation efficiency.
Empirical moderation tests using firm-level panel data (2012–2023) showing interaction between AI application measures and fiscal support intensity.
Market segmentation exerts a moderating effect on the relationship between AI application and enterprise innovation efficiency.
Moderation analysis in the empirical framework applied to the 2012–2023 panel of Shanghai and Shenzhen A-share firms showing interaction effects between AI application and market segmentation measures.
The utility-aware framework preserves inverse U-shaped demand patterns for attributes such as aesthetics and uniqueness, improving demand-based performance while preserving fidelity and semantic consistency.
Empirical claim from the paper that their method maintains observed inverse U-shaped demand relationships for certain attributes in their experiments while improving demand-related metrics.
Modern retrieval agents expose many configuration choices -- LLM, retriever, number of documents, number of hops, and synthesis strategy -- each shaping both answer quality and serving cost.
Paper's conceptual description of retrieval pipelines and configuration dimensions (LLM, retriever, number of documents, number of hops, synthesis strategy). No empirical sample size reported for this descriptive claim.
AI's future impact on employment will depend not only on automation capabilities but also on how responsibly enterprises manage workforce transitions.
Paper's concluding claim synthesizing arguments and proposed governance approach (normative conclusion rather than an empirically tested causal estimate in the excerpt).
AI-induced workforce disruption is not only a labor market issue but also an enterprise governance challenge.
Argument/position advanced in the paper highlighting governance responsibilities for firms implementing AI.
Artificial intelligence, especially generative AI, is transforming enterprise operations by automating tasks, enhancing decision-making, and redefining job roles.
Conceptual statement in the paper describing observed/expected effects of generative AI on enterprise operations (no specific empirical sample or experiment reported in the excerpt).
Depending on operational parameters, the most time-efficient way to complete a workflow may undergo a transition between two task-processing regimes: a fully AI-assisted regime and a fully manual regime.
Analytical results derived from the paper's formal queueing model (theoretical/model-based derivation; no empirical sample reported).
AI assistance can generate a deceptive productivity signature: average completion times fall because AI tools typically supply a fast first draft, yet workflow-level performance can deteriorate when a subset of AI errors escapes review and returns as costly downstream rework.
Analytical derivation and discussion based on the paper's queueing model (theoretical/model-based evidence; no empirical sample provided).
Drawing on the partial equilibrium model of Gries and Naudé (2022), existing economic frameworks may inadvertently overlook these factors.
The paper's theoretical critique referencing Gries & Naudé (2022); argument is based on model comparison and conceptual analysis rather than new empirical tests.
We identify five key moderating factors: human resource composition, baseline capability of individuals, learning curve of practitioners, incentives for fair use, and flexibility of objectives.
Explicit enumeration of proposed moderating factors in the paper (conceptual identification rather than empirical measurement).
Following the advent of high-performance generative models, AI use has been rapidly encouraged in some sectors while being restricted in others.
Descriptive claim in the paper's introduction/abstract; based on observation and literature context rather than new empirical data.
The framework does not force domains into the same shape; it surfaces each domain's actuarial geometry.
Empirical observation of differing frontier shapes and capital demands across the instantiated domains and traces.
Required reserve capital varies by 22x (Capital@50 from 289 to 6457).
Quantitative results reported in experiments across domains (Capital@50 values reported for domains; ratio computed).
The frontier exhibits a common low-reserve refusal and intermediate-release pattern across domains, with saturation only where the budget grid reaches full reserve demand.
Observed pattern reported across the four instantiated environments and the retail/airline tau-bench traces in experimental results.
AI can raise productivity and output, but its distributional effects are uncertain and mediated by institutions and access to complementary resources.
Conceptual claim in abstract synthesizing literature; supported by secondary sources and integrative framework (OECD, ILO, UNDP, WTO, WEF). No quantified sample size reported.
These findings have broader implications for productivity, equity, and capacity across the global research system.
Discussion/interpretation in paper based on causal results from randomized experiment; inference from observed behavioral changes and heterogeneous effects.
Completion time itself is not sufficient to characterize efficiency gains.
Authors' inferential conclusion in the abstract based on observed dissociation between completion time (no difference) and subjective effort (lower with AI) in their preregistered study (N = 1237).
Decomposition analysis reveals that wage benefits are concentrated among employees aged 45 and above, managers, and white-collar workers; other worker categories experience stagnant wages, and no group shows a negative wage effect.
Decomposition of wage effects by worker groups (age, occupation/type) using the integrated dataset and the DiD/other regression analyses.
Wage increases at small firms primarily explain the positive adoption effect, while wages at medium and large firms remain stagnant after adoption.
Heterogeneity analysis by firm size within the DiD framework showing differential post-adoption wage trajectories for small versus medium/large firms.
Key mechanisms of AI's impact on employment structure were identified: automation of routine processes, formation of new professional profiles, and changes in requirements for employees' competencies.
Qualitative analysis of statistical data, industry reviews, and regulatory legal documents described in the paper (no experimental or survey sample size reported).
We propose the Shannon Scaling Law, a unified theoretical framework that models LLM training as information transmission over a noisy channel, grounded in the Shannon-Hartley theorem, mapping model parameters to channel bandwidth and training tokens to signal power.
Theoretical formulation presented in the paper, grounded on Shannon-Hartley theorem and a mapping between model/data quantities and communication-theoretic quantities (bandwidth, signal power).
Models with near-identical overall strength show qualitatively different capability profiles.
Observed differences in capability-profile axes for models with similar aggregate scores in the tournament.
Managerial traits, such as risk tolerance and patience, play a role in shaping firms' AI adoption decisions.
Inclusion of manager-level trait measures (risk tolerance, patience) in the ifo Business Survey and analysis showing associations between these traits and reported AI adoption.
Drivers and barriers to AI adoption include firm-specific characteristics and industry dynamics.
Survey-based analysis linking firm characteristics and industry-level factors to reported AI adoption decisions in the ifo Business Survey (likely correlational/regression analysis).
AI adoption/diffusion varies across firm sizes.
Analysis of adoption patterns by firm size using ifo Business Survey firm-level responses (comparison across size categories).
AI is changing informal cultural practices like professional mentoring that are key to helping professionals settle in their positions, stay engaged with their work, and grow their careers.
Participant reports from the 24 interviews indicating changes to informal practices such as mentoring, onboarding, and informal feedback.
AI is changing formal role responsibilities and collaborations between those roles.
Qualitative interview data from 24 product-focused employees describing shifts in formal responsibilities and inter-role collaboration.
AI adoption is allowing professionals to blur and extend the boundaries of their corporate roles.
Reported by interview participants (qualitative evidence) from the 24 interviews at one large technology firm.
Artificial Intelligence (AI) has caused massive changes in nature of workplaces in healthcare sector.
Asserted in paper's introduction and supported by a scoping review (PRISMA-ScR) of 29 peer-reviewed empirical studies published 2020–2025.
The urban digital economy exerts a stronger effect than the rural digital economy in promoting servicization and inhibiting industrialization.
Heterogeneity analysis in the provincial panel (2013–2024) comparing urban versus rural digital-economy measures and their associations with changes in employment shares.
After 2017, industrial digitalization continued to strengthen servicization while suppressing industrialization.
Post-2017 analysis of provincial panel data (2013–2024) showing continued positive association of industrial digitalization with service employment and negative association with industrial employment after 2017.
After 2017, digital industrialization shifted toward promoting industrialization and restraining servicization.
Post-2017 subset analysis of provincial panel data (2013–2024) comparing the direction and magnitude of digital industrialization's association with industry and service employment shares before and after 2017.
The elevation of the 'digital economy' to a national strategy in 2017 constituted a critical turning point in the relationship between digital-economy development and labor-structure change.
Before-and-after (pre/post-2017) analysis using China's provincial panel data (2013–2024) showing a structural change in estimated effects around 2017.
The development of the digital economy generally promotes the servicization and deindustrialization of the labor structure.
Panel analysis using China's provincial data from 2013 to 2024 examining relationships between digital economy development and labor-structure indicators (servicization and industrial employment shares).
Benchmark-based evaluation can both overstate and understate deployed capability because it privileges tasks that can be precisely specified, automatically graded, easy to optimize for, and run with low budgets and short time horizons.
Analytical argument in the paper (theoretical/qualitative critique of benchmark methodology); supported by a survey of recent open-world evaluations (method description in paper), but no quantified cross-benchmark empirical study reported in the abstract.