Evidence (7156 claims)
Adoption
5126 claims
Productivity
4409 claims
Governance
4049 claims
Human-AI Collaboration
2954 claims
Labor Markets
2432 claims
Org Design
2273 claims
Innovation
2215 claims
Skills & Training
1902 claims
Inequality
1286 claims
Evidence Matrix
Claim counts by outcome category and direction of finding.
| Outcome | Positive | Negative | Mixed | Null | Total |
|---|---|---|---|---|---|
| Other | 369 | 105 | 58 | 432 | 972 |
| Governance & Regulation | 365 | 171 | 113 | 54 | 713 |
| Research Productivity | 229 | 95 | 33 | 294 | 655 |
| Organizational Efficiency | 354 | 82 | 58 | 34 | 531 |
| Technology Adoption Rate | 277 | 115 | 63 | 27 | 486 |
| Firm Productivity | 273 | 33 | 68 | 10 | 389 |
| AI Safety & Ethics | 112 | 177 | 43 | 24 | 358 |
| Output Quality | 228 | 61 | 23 | 25 | 337 |
| Market Structure | 105 | 118 | 81 | 14 | 323 |
| Decision Quality | 154 | 68 | 33 | 17 | 275 |
| Employment Level | 68 | 32 | 74 | 8 | 184 |
| Fiscal & Macroeconomic | 74 | 52 | 32 | 21 | 183 |
| Skill Acquisition | 85 | 31 | 38 | 9 | 163 |
| Firm Revenue | 96 | 30 | 22 | — | 148 |
| Innovation Output | 100 | 11 | 20 | 11 | 143 |
| Consumer Welfare | 66 | 29 | 35 | 7 | 137 |
| Regulatory Compliance | 51 | 61 | 13 | 3 | 128 |
| Inequality Measures | 24 | 66 | 31 | 4 | 125 |
| Task Allocation | 64 | 6 | 28 | 6 | 104 |
| Error Rate | 42 | 47 | 6 | — | 95 |
| Training Effectiveness | 55 | 12 | 10 | 16 | 93 |
| Worker Satisfaction | 42 | 32 | 11 | 6 | 91 |
| Task Completion Time | 71 | 5 | 3 | 1 | 80 |
| Wages & Compensation | 38 | 13 | 19 | 4 | 74 |
| Team Performance | 41 | 8 | 15 | 7 | 72 |
| Hiring & Recruitment | 39 | 4 | 6 | 3 | 52 |
| Automation Exposure | 17 | 15 | 9 | 5 | 46 |
| Job Displacement | 5 | 28 | 12 | — | 45 |
| Social Protection | 18 | 8 | 6 | 1 | 33 |
| Developer Productivity | 25 | 1 | 2 | 1 | 29 |
| Worker Turnover | 10 | 12 | — | 3 | 25 |
| Creative Output | 15 | 5 | 3 | 1 | 24 |
| Skill Obsolescence | 3 | 18 | 2 | — | 23 |
| Labor Share of Income | 7 | 4 | 9 | — | 20 |
AI appears to be a diffusing technology, not an emerging occupation.
Synthesis of empirical findings: presence of a shared vocabulary but lack of a coherent practitioner population in resume data, interpreted as diffusion of AI skills/vocabulary across existing roles.
Across heterogeneous learners, a common broadcast curriculum can be slower than personalized instruction by a factor linear in the number of learner types.
Theoretical comparative result in the model (analysis of broadcast vs personalized curricula across heterogeneous learner types; abstract states factor linear in number of types).
The findings provide evidence against cue-based accounts of lie detection more generally.
Authors' interpretation: because lie-detection accuracy did not decrease despite changes to visual cues (retouching, backgrounds, avatars), the results challenge theories that rely on superficial cues for lie detection.
Participants' confidence in their judgments declined in AI-mediated videos, particularly when some participants used avatars while others did not.
Experimental comparisons across conditions with varying levels of AI mediation; subgroup/condition contrast highlighting larger declines in mixed-avatar settings.
Perceived trust in speakers declined in AI-mediated videos.
Experimental results from the two preregistered online experiments comparing perceived trust across varying levels of AI mediation (retouching, background replacement, avatars).
AI-based tools that mediate, enhance or generate parts of video communication may interfere with how people evaluate trustworthiness and credibility.
Motivating claim stated in the paper's introduction/abstract; not an empirical finding but a hypothesis motivating the experiments.
AI adoption faces critical obstacles originating from digital illiteracy, poor Internet access, excessive application costs, and the rural-to-urban divide.
Survey findings and interview themes from the mixed-methods study (survey n=293; interviews n=12) identifying barriers to AI adoption.
Users still had concerns about how AI credit assessments and chatbots operate.
Qualitative interview data (n=12) and/or survey responses (n=293) reporting user concerns about AI credit scoring and chatbots.
Compositional spatial reasoning remains a formidable challenge for state-of-the-art VLMs (as revealed by our evaluation).
Empirical results from the evaluation of the 37 VLMs on the MultihopSpatial benchmark showing poor performance on multi-hop/compositional queries.
Existing benchmarks predominantly focus on elementary, single-hop relations and neglect multi-hop compositional spatial reasoning and precise visual grounding needed for real-world scenarios.
Literature/benchmark survey and motivation presented by the authors comparing characteristics of prior benchmarks vs. the proposed needs.
Adoption barriers exist, particularly for small and medium-sized enterprises and firms in emerging economies, where capability and data constraints limit impact.
Findings reported from the systematic review and mixed-methods assessment (abstract references barriers observed across reviewed studies); number of studies reported in abstract is 104 for the systematic review.
AI can initially exacerbate distributional injustice.
Dimension-level analysis indicating negative (or initially negative) effects of AI on the distributional component of the energy justice index.
There are few integrated frameworks (bridging ethics and technical controls) in the current AI governance landscape.
Result of the literature review and cluster analysis showing limited coverage of frameworks that integrate ethical principles with auditable technical controls.
Findings reveal a fragmented landscape dominated by ethics/privacy-centric and compliance/risk-focused approaches.
Synthesis of the reviewed literature and results of PCA/k-means clustering indicate thematic dominance of ethics/privacy and compliance/risk orientations across frameworks.
Significant limitations emerged in case law citations, with most cited cases being non-existent or incorrectly referenced.
Authors' review of the case citations produced by the four AI engines for the single transcript, finding many citations were fabricated or misreferenced.
These findings uncover critical threats to judicial integrity and public trust and underscore the urgent need for robust safeguards against non-legal influences in AI legal systems.
Interpretation/conclusion drawn from the empirical results (observed deviations, sentiment amplification, and subgroup vulnerabilities).
These safety risks are compounded for emotionally charged topics.
Subgroup analyses where emotionally charged case topics showed larger deviations and stronger effects from injected sentiment.
These safety risks are compounded (stronger) for low-skilled occupational categories.
Subgroup analyses reported in the paper showing larger model deviations and/or greater sentiment amplification effects for cases involving low-skilled occupations.
The sentiment-induced divergences lead to unstable and often inflated compensation predictions by the models.
Analysis of model-predicted compensation amounts under sentiment perturbations showing increased variability and upward bias compared to CJOL amounts.
Public opinion (social media sentiment) substantially amplifies deviations between LLM outputs and real rulings.
Stress-test experiments in which injected social media sentiment increased the divergence of model outputs from CJOL judgments across the sample.
Models exhibit inherent deviations from real rulings.
Empirical comparison of LLM outputs to CJOL judgments showing systematic differences (based on the paper's reported comparisons across the dataset).
GDP growth is initially negatively affected by the ageing population.
Estimated negative association reported in panel threshold regressions using provincial panel data (31 provinces, 2000–2022); ageing operationalized (primary specification) as an ageing measure (paper also tests old-age dependency ratio).
The article argues that the idea of a “Pax Silica” is fragile.
Conclusion drawn from the paper's theoretical framework and comparative analysis; presented as an assessment rather than empirical measurement.
Contemporary struggles over semiconductor supply chains represent not a new hegemonic order but a logistical adaptation of Pax Americana.
Stated thesis supported by comparative/historical analysis and theoretical argumentation (comparative analysis of historical Pax orders and U.S. techno-security architecture); no quantitative sample size reported in abstract.
Initial adaptation challenges to AI integration were identified among employees.
Participants in semi-structured interviews (n=12) reported initial difficulties adapting to AI tools; themes relating to early adaptation challenges were coded.
Past machine learning applications to pricing have produced models that adapt slowly to real-time changes, depend heavily on historical data, and struggle to handle multi-agent scenarios.
Stated as literature/related-work critique in paper; no new empirical evidence or sample size provided in the excerpt.
Traditional methods, such as rule-based algorithms and statistical scale forecasting, struggle to adapt to rapidly changing market conditions, competitive maneuvers, and evolving consumer strategies, leading to sub-optimal pricing and decreased profitability.
Paper asserts this as background/motivation; no detailed empirical study or sample size provided in the excerpt.
In the short term, big data may inhibit welfare growth.
Theoretical comparative-static/dynamic analysis reported in the model showing that initial or short-run effects of increased data sharing can reduce welfare growth (no empirical/sample data).
There is a measurement asymmetry in standard LLM evaluation: unconstrained prompts can inflate constraint-adherence scores and mask the practical value of structured prompting.
Analysis of evaluation results from the controlled study showing that unconstrained (simple) prompts sometimes achieve high constraint-adherence scores, leading to misleading evaluation of structured prompts' benefits.
Traditional paradigms, specifically the resource-based view and the dynamic capabilities framework, operate under closed-system, first-order cybernetic assumptions that fail to capture the dissipative nature of algorithmic agents.
Conceptual critique presented in the paper's theoretical argumentation (literature critique and re-framing); no empirical sample reported.
AI usage predicts work disengagement behavior via emotional exhaustion elicited by AI-associated technostressors.
Four-stage longitudinal study (survey) of finance professionals (N=285); mediation analysis testing AI usage -> technostressors -> emotional exhaustion -> work disengagement, based on SOR framework.
These findings highlight fundamental challenges in the numerical and time-series reasoning for current LLMs and motivate future research in financial intelligence.
Interpretation of experimental results in the paper: authors conclude that the observed limited gains (particularly on trading-signal/time-series aspects) indicate shortcomings in LLM numerical and time-series reasoning.
There is a central design tension in human-AI systems: maximizing short-term hybrid capability does not necessarily preserve long-term human cognitive competence.
Conceptual/theoretical claim derived from the framework and discussion in the paper (argument and mathematical framing), no empirical sample or longitudinal data presented in the excerpt.
This result directly contradicts classical scaling laws which assume monotonic capability gains with model scale.
Comparative theoretical claim in the paper contrasting the Institutional Scaling Law with classical empirical/theoretical scaling laws in ML literature.
The Institutional Scaling Law proves that institutional fitness is non-monotonic in model scale.
Formal mathematical derivation/proof presented in the paper (the 'Institutional Scaling Law').
AI development proceeds not through smooth advancement but through extended periods of stasis interrupted by rapid phase transitions that reorganize the competitive landscape (punctuated equilibrium pattern).
Argument based on punctuated equilibrium theory from evolutionary biology and historical analysis presented in the paper identifying discrete transitions in AI history; the paper cites and classifies eras/events as evidence.
The interaction of artificial intelligence and environmental regulation produces a '1 + 1 < 2' crowding-out effect (their combined effect is less than the sum of individual effects).
Spatial Durbin model with interaction term between AI and environmental regulation as summarized in the abstract; reported as a crowding-out interaction.
Environmental regulation significantly inhibits local UCEE.
Spatial Durbin model results reported in the abstract indicating a significant negative local coefficient for environmental regulation.
Artificial intelligence significantly inhibits local UCEE.
Spatial Durbin model results reported in the abstract indicating a significant negative local coefficient for artificial intelligence.
Progress in agentic AI systems that generate and optimize GPU kernels is constrained by benchmarks that reward speedup over software baselines rather than proximity to hardware-efficient execution.
Author argument/observation in paper (conceptual claim about limitations of existing benchmarks); no empirical sample or experiment reported in the provided text.
Rather than broad job losses, evidence points to a reallocation at the entry level: AI automates tasks typically assigned to junior staff, shifting the nature of entry-level roles.
Synthesis of firm- and task-level empirical studies reported in the brief documenting automation of routine/junior tasks and changes in job-task composition; specific sample sizes vary by cited study and are not provided in the brief.
Algorithmic credit systems are linked to higher levels of financial stress.
Study reports a positive association between algorithmic credit system use and reported financial stress from regression analysis on the 400-user cross-sectional dataset.
Confirmation bias poses a weakness in LLM-based code review, with implications on how AI-assisted development tools are deployed.
Synthesis of findings from Study 1 (framing-induced detection failures) and Study 2 (practical exploitability and partial mitigation via debiasing).
Adversarial framing succeeds in 88% of cases against Claude Code (autonomous agent) in real project configurations where adversaries can iteratively refine their framing to increase attack success.
Study 2 experiments in real project configurations with iterative adversary refinement evaluated against Claude Code (autonomous agent); reported 88% success rate.
Adversarial pull request framing (e.g., labeled as security improvements or urgent functionality fixes) succeeds in reintroducing known vulnerabilities in 35% of cases against GitHub Copilot under one-shot attacks.
Study 2 experiments simulating adversarial pull requests evaluated against GitHub Copilot (interactive assistant); reported success rate 35% for one-shot attacks.
The framing effect is strongly asymmetric: false negatives increase sharply while false positive rates change little.
Comparison of false negative and false positive rates across framing conditions in Study 1 experiments (250 CVE pairs across models).
Framing a change as bug-free reduces vulnerability detection rates by 16-93%.
Result reported from Study 1 controlled experiments across models and framing conditions (250 CVE pairs).
AI-only baselines perform near or below the median of competition participants.
Comparison of AI-only baseline performance to the distribution of competition participant results reported in the paper (competition with 29 teams / 80 participants).
Our results show that current AI agents struggle with domain-specific reasoning.
Outcome of the competition reported in the paper comparing AI-only baselines to participant submissions across the AgentDS tasks (competition data from 29 teams / 80 participants); reported aggregate performance indicating AI weakness on domain-specific tasks.
LLM-generated peer reviews place significantly less weight on clarity and significance of the research.
Comparative analysis between LLM-generated reviews and human reviews from the conference dataset; reported as a statistically significant difference but exact statistics and sample size not provided in the excerpt.