Evidence (13870 claims)
Adoption
8467 claims
Productivity
7558 claims
Governance
6805 claims
Human-AI Collaboration
6363 claims
Org Design
4132 claims
Innovation
4065 claims
Labor Markets
3526 claims
Skills & Training
2945 claims
Inequality
2066 claims
Evidence Matrix
Claim counts by outcome category and direction of finding.
| Outcome | Positive | Negative | Mixed | Null | Total |
|---|---|---|---|---|---|
| Other | 749 | 196 | 98 | 892 | 1984 |
| Governance & Regulation | 817 | 394 | 188 | 121 | 1544 |
| Organizational Efficiency | 771 | 189 | 124 | 83 | 1177 |
| Technology Adoption Rate | 627 | 233 | 123 | 96 | 1088 |
| Research Productivity | 411 | 123 | 56 | 332 | 933 |
| Output Quality | 467 | 178 | 59 | 47 | 751 |
| Decision Quality | 320 | 174 | 75 | 42 | 618 |
| Firm Productivity | 435 | 55 | 88 | 20 | 604 |
| AI Safety & Ethics | 214 | 276 | 65 | 33 | 593 |
| Market Structure | 178 | 167 | 122 | 24 | 496 |
| Task Allocation | 207 | 64 | 71 | 32 | 379 |
| Skill Acquisition | 165 | 59 | 60 | 17 | 301 |
| Innovation Output | 203 | 27 | 43 | 18 | 292 |
| Employment Level | 105 | 52 | 107 | 13 | 279 |
| Fiscal & Macroeconomic | 131 | 69 | 43 | 26 | 276 |
| Consumer Welfare | 116 | 63 | 42 | 11 | 232 |
| Firm Revenue | 150 | 48 | 26 | 3 | 227 |
| Inequality Measures | 44 | 122 | 49 | 6 | 221 |
| Task Completion Time | 169 | 29 | 8 | 12 | 219 |
| Worker Satisfaction | 89 | 63 | 20 | 12 | 184 |
| Error Rate | 69 | 92 | 10 | 2 | 173 |
| Regulatory Compliance | 76 | 68 | 14 | 5 | 163 |
| Training Effectiveness | 93 | 21 | 13 | 19 | 148 |
| Wages & Compensation | 77 | 36 | 25 | 6 | 144 |
| Automation Exposure | 51 | 54 | 22 | 12 | 142 |
| Team Performance | 86 | 17 | 27 | 9 | 140 |
| Developer Productivity | 94 | 17 | 14 | 6 | 132 |
| Job Displacement | 12 | 80 | 20 | 1 | 113 |
| Hiring & Recruitment | 51 | 7 | 8 | 3 | 69 |
| Creative Output | 31 | 17 | 7 | 3 | 59 |
| Skill Obsolescence | 5 | 46 | 6 | 1 | 58 |
| Social Protection | 27 | 16 | 8 | 2 | 53 |
| Labor Share of Income | 17 | 17 | 17 | — | 51 |
| Worker Turnover | 11 | 12 | — | 3 | 26 |
| Industry | — | — | — | 1 | 1 |
This work offers a cost-effective, scientifically grounded blueprint for ubiquitous AI education.
Authors' concluding statement based on the SOP, low labor/hardware claims, and the pilot exam results showing high accuracy with the Shadow Agent in newer 32B models.
This suggests that structured reasoning guidance (as implemented by the Shadow Agent) is the key to unlocking the latent power of modern small language models.
Interpretive claim based on the pilot study's observed large gains for newer 32B models when using Shadow Agent guidance versus smaller gains for older models and stagnation in baselines.
In contrast, older models see only modest gains (~10%) from the Shadow Agent guidance.
Same pilot study reporting that older (unspecified) model generations showed only about a ~10% improvement when using the Shadow Agent versus baseline. No exact accuracy numbers, sample size, or model names provided.
The Shadow Agent, which provides structured reasoning guidance, triggers a massive capability surge in newer 32B models, boosting performance from 74% (Naive RAG) to mastery level (90%).
Pilot study on a full graduate-level final exam reported comparisons between Naive RAG (74% accuracy) and the Shadow Agent (90% accuracy) for newer 32B models. Specific number of exam items or statistical testing not stated.
We used a Vision-Language Model data cleaning strategy and a novel Shadow-RAG architecture as core technical components of the localization pipeline.
Methodological description in the practitioner report; the paper explicitly names these two techniques as the data-cleaning and architectural contributions used to create the tutor.
Using a Vision-Language Model data cleaning strategy and a novel Shadow-RAG architecture, we localized a graduate-level Applied Mathematics tutor using only 3 person-days of non-expert labor and open-weights 32B models deployable on a single consumer-grade GPU.
Practitioner report describing a replicable Standard Operating Procedure (SOP); method claims include Vision-Language Model data cleaning and Shadow-RAG; deployment described as using open-weight 32B models on a single consumer GPU; labor reported as '3 person-days of non-expert labor'. No sample size or independent replication reported in text.
If you can prove the value and the effort behind API token spending (agent memory), you can resell it.
Normative/operational claim within the paper's proposal; presented as an implication of verifiable provenance and market layering, with no empirical proof or transactional data.
Enabling timely memory transfer reduces repeated exploration.
Argument in the paper asserting that shared/tradable memory decreases redundant exploration; no experimental or observational data provided.
Together, clawgang and meowtrade transform one-shot API token spending into reusable and tradable assets.
High-level systems argument in the paper; no empirical measurements of reuse or tradability presented.
Meowtrade is a market layer for listing, transferring, and governing certified memory artifacts.
Design proposal described in the paper; no pilot deployment, user adoption metrics, or experimental data provided.
Clawgang binds memory to verifiable computational provenance.
System/design claim describing the proposed mechanism (clawgang) in the paper; no implementation results or empirical validation reported.
Agent memory can serve as an economic commodity in the agent economy, if buyers can verify that it is authentic, effort-backed, and produced in a compatible execution context.
Conceptual argument in the paper's proposal; no empirical evaluation, sample size, or experiments reported.
Economic theory can be used to generate structured synthetic data that improves foundation-model predictions when the theory implies observable patterns in the data.
General conclusion drawn from the paper's experimental findings: improvement in model predictions after fine-tuning on theory-derived synthetic data.
Fine-tuning on GARP-consistent synthetic data substantially improves prediction relative to zero-shot Chronos-2 at all forecast horizons we study.
Empirical results comparing fine-tuned Chronos-2 to zero-shot Chronos-2 across multiple forecast horizons on the authors' experimental panel (no numeric metrics or sample sizes given in the excerpt).
The fine-tuned model serves as a rationality-constrained forecasting prior: it learns price-quantity relations from GARP-consistent synthetic histories and then uses those relations to predict the choices of real consumers.
Empirical approach described in paper: model fine-tuned on synthetic GARP-consistent histories and then evaluated on real consumer choice data (supports claim that model transfers learned relations to predicting real choices).
GARP is a simple condition to check that allows us to generate time series from a large class of utilities efficiently.
Methodological argument in the paper: authors use GARP as a constructive condition to generate synthetic time series from many utility functions (no numeric efficiency metrics provided in the excerpt).
Teaching them basic economic logic improves how they predict demand using an experimental panel.
Reported experimental results in the paper: fine-tuning models on synthetic, economics-consistent data and evaluating on an experimental panel of consumer demand (no numeric sample size or metrics provided in the excerpt).
AI adoption and the associated improved governance lead to higher total factor productivity (TFP).
Empirical analysis showing a positive association between firm-level AI application index and measures of total factor productivity in the 2010–2023 Chinese A-share panel.
AI adoption and the associated improved governance lead to a lower cost of debt financing for firms.
Empirical tests linking firm-level AI application and governance improvements to measures of debt financing costs (e.g., interest rates on debt, financing spreads) in the Chinese A-share firm sample.
The governance risk-mitigation effects of AI operate through enhancing external monitoring.
Mechanism analyses showing that AI adoption is associated with measures of stronger external monitoring (e.g., analyst coverage, media scrutiny, regulator activity) in the firm-year panel, linking that channel to reduced misconduct.
The governance risk-mitigation effects of AI operate through strengthening internal control capacity.
Mechanism analyses showing that higher AI application is associated with improved internal control measures (as reported by firms or regulatory/financial-control indicators) in the dataset of Chinese A-share firms.
The governance risk-mitigation effects of AI operate through lowering agency costs.
Mechanism analyses reported by authors linking AI adoption to reductions in measures interpreted as agency costs (e.g., agency-cost proxies, corporate governance metrics) in the same firm-year panel.
AI application significantly reduces the monetary amount of penalties associated with executive misconduct.
Regression analyses on monetary penalty data for Chinese A-share firms (2010–2023) showing a statistically significant negative relationship between firm AI application index and penalty amounts.
AI application significantly reduces the frequency (number) of violations by executives.
Empirical frequency/regression analyses on the firm-year panel of Chinese A-share firms using the AI application index; authors report robust reductions in the number/frequency of violations conditional on AI adoption.
AI application significantly reduces the incidence of executive misconduct.
Empirical analysis on Chinese A-share listed firms (2010–2023) using the constructed firm-level AI application index; reported significant negative association between AI application and whether a firm experiences executive misconduct (incidence).
Using Chinese A-share firms listed in Shanghai and Shenzhen from 2010 to 2023, we construct a firm-level AI application index and examine whether and how AI adoption mitigates executive misconduct.
Authors report building a firm-level AI application index and applying it to Chinese A-share listed firms (Shanghai and Shenzhen) over 2010–2023 to study links between AI adoption and executive misconduct (method: panel analysis using firm-year observations).
Applying our framework to product listings on Etsy, we find that following ChatGPT's release, listings have significantly more machine-usable information about product selection, consistent with systematic mecha-nudging.
Empirical analysis of Etsy product listings comparing measures of 'machine-usable information about product selection' before and after ChatGPT's release. (The abstract states a significant increase; full paper presumably contains dataset details and statistical tests, but sample size and exact estimates are not provided in the excerpt.)
Adoption of AI can reduce procurement costs by 15.7%.
Field survey data (n=326) and regression analysis; authors report a 15.7% reduction in procurement costs associated with AI adoption.
Adoption of AI can shorten the procurement decision-making cycle by 21.3%.
Field survey data (n=326) analyzed (authors report a 21.3% reduction in procurement decision-making cycle associated with AI adoption); method described as questionnaire surveys and multiple linear regression.
Supplier AI capability positively drives AI adoption in procurement (β = 0.28, p < 0.01).
Same questionnaire survey (n=326) and multiple linear regression analysis; reported coefficient β=0.28 with p<0.01.
Perceived usefulness positively drives AI adoption in procurement (β = 0.32, p < 0.01).
Questionnaire survey of 326 procurement managers/supply chain managers in SMEs (Yangtze River Delta and Pearl River Delta) analyzed using multiple linear regression; reported coefficient β=0.32 with p<0.01.
The paper provides recommendations for designing strategic indicators to drive adoption, foster innovation, and objectively assess whether digital tools are delivering top-line impact.
Descriptive claim about the content of the perspective article (the authors state they provide these recommendations); the excerpt itself summarizes this contribution.
The shift from expert-driven computer-aided drug design (CADD) to semiautonomous AI necessitates a new framework of impact-oriented KPIs.
Stated by the EFMC2 community authors as a normative conclusion in the perspective piece; based on the characterisation of a technological shift rather than on presented empirical tests in the excerpt.
Harnessing AI's potential requires moving beyond measuring technical model performance (e.g., predictive accuracy) to measuring strategic impact.
Authors argue this as a conceptual requirement for realizing AI's benefits in R&D; presented as a recommendation rather than supported by quantified empirical evidence in the excerpt.
Preliminary analyses suggest that 'AI-native' companies may be outpacing traditional peers.
Explicitly stated in the paper as based on preliminary analyses; the excerpt provides no details on the analyses, metrics, or sample sizes.
The broad introduction of AI into the R&D landscape over the last years holds the promise to lift pharmaceutical R&D out of its productivity problem.
Framed as an expectation/promise in the paper; based on recent broad adoption trends of AI in R&D (no specific empirical evaluation or sample size reported in the excerpt).
The visualization preserved human control.
Reported result from the within-subjects experiment (N=32) indicating that using the visualization did not reduce human control/agency in the negotiation process.
In the same within-subjects experiment (N=32), the visualization improved efficiency.
Within-subjects experiment (N=32) reported in the paper; the authors state the visualization improved efficiency (likely measured as time, number of rounds, or steps to reach agreement).
In a within-subjects experiment (N=32), the uncertainty-based visualization improved human outcomes.
Within-subjects user experiment reported in the paper with N=32 participants comparing performance with and without the visualization.
We introduce a novel uncertainty-based visualization driven by Bayesian estimation of agreement probability that shows how the space of mutually acceptable agreements narrows as negotiation progresses, helping users identify promising options.
Design and implementation of a visualization technique described in the paper; the visualization is driven by Bayesian estimation of agreement probability and is presented as a tool to reveal the shrinking feasible agreement space during negotiation.
In this verifiable domain, simple arbitrage strategies generate net profit margins of up to 40%.
Empirical result from the SWE-bench case study comparing arbitrage strategy returns using GPT-5 mini and DeepSeek v3.2 (reported maximum net profit margin = 40%).
Generative AI can autonomously produce novel content, including text, images, models, and scenarios.
General technical/descriptive claim stated in the paper's background/introduction; not an empirically tested claim within the provided excerpt.
Generative AI facilitates the synthesis of structured and unstructured information from diverse sources, enabling managers to explore multiple decision pathways, identify potential risks, and optimize strategic choices.
Descriptive/functional claim made in the paper's introduction and conceptual framing; the empirical component (survey + SEM) is described generally but no specific measures or effect sizes for information synthesis or these capabilities are provided in the excerpt.
Generative AI augments human creativity by producing innovative solutions and scenario-planning alternatives that may not emerge through conventional analytical approaches.
Stated in the conceptual/argumentative portion of the paper; may be supported by survey items but no explicit empirical measure or effect size for creativity is provided in the provided text.
Decision quality and strategic agility positively influence organizational performance.
Reported SEM results from the paper linking the constructs (decision quality and strategic agility) to organizational performance using survey data from senior managers and AI adoption specialists; method = SmartPLS.
Generative AI adoption significantly enhances strategic agility.
Same empirical source as above: survey of senior managers/decision-makers/AI adoption specialists; tested via Structural Equation Modeling (SmartPLS) as reported in the paper.
Generative AI adoption significantly enhances decision quality.
Empirical analysis reported in the paper: survey data collected from senior managers, decision-makers, and AI adoption specialists across multiple industries; relationships assessed using Structural Equation Modeling (SmartPLS). No numeric sample size or effect estimate reported in the provided text.
Human-like presentations increased perceived usefulness and agency in certain tasks.
Experimental manipulation of the human-likeness of AI presentation in the study's three tasks; the abstract reports increased perceived usefulness and agency for human-like presentations in some tasks. No sample sizes, task specifics, or effect magnitudes reported in abstract.
A single dissent within a panel reduced pressure to conform.
Experimental manipulation of within-panel consensus (introducing a single dissent) in the study's three tasks; abstract reports that a single dissent lowered conformity pressure. No numerical data provided in abstract.
Accuracy improved for small panels relative to a single AI.
Reported experimental result from the paper's study: participants completed three tasks and received advice from AI panels; panel size was manipulated (small panels vs single AI). The abstract states this accuracy improvement for small panels. (Sample size and exact tasks not reported in abstract.)