Evidence (7560 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
Filtered →
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 |
Human Ai Collab
Remove filter
These negative effects (reduced persistence and impaired unassisted performance) emerge after only brief interactions with AI (approximately 10 minutes).
Experimental manipulation / exposure in RCTs where participants interacted with AI for about 10 minutes and subsequent outcomes were measured.
People are more likely to give up after interacting with AI (increased likelihood of quitting tasks unassisted).
Randomized controlled trials (N = 1,222) measuring rates of task abandonment/giving-up after AI interaction vs. control.
AI assistance impairs unassisted performance: although AI improves short-term performance, people perform significantly worse without AI after interacting with it.
Randomized controlled trials (N = 1,222) comparing performance with and without AI assistance across tasks; causal inference from randomized assignment.
Through a series of randomized controlled trials on human-AI interactions (N = 1,222), we provide causal evidence that AI assistance reduces persistence.
Randomized controlled trials (RCTs) on human-AI interactions with total sample size N = 1,222; persistence measured after AI interaction across tasks.
AI-assisted evaluation reduces variance in research quality.
SEM and regression analyses on OECD panel data report a decrease in variance of research quality measures associated with higher AIRC.
High-risk agentic systems with untraceable behavioral drift cannot currently satisfy the AI Act's essential requirements.
Authors' legal and normative conclusion based on their regulatory mapping and analysis (argumentative/legal reasoning rather than reported empirical testing).
The paper identifies agent-specific compliance challenges in cybersecurity, human oversight, transparency across multi-party action chains, and runtime behavioral drift.
Author-stated findings from the regulatory mapping and analysis; specific challenge areas listed without reported quantitative measurement.
The EU AI Act (Regulation 2024/1689) regulates these systems through a risk-based framework, but it does not operate in isolation: providers face simultaneous obligations under the GDPR, the Cyber Resilience Act, the Digital Services Act, the Data Act, the Data Governance Act, sector-specific legislation, the NIS2 Directive, and the revised Product Liability Directive.
Legal/regulatory mapping asserted by the authors listing specific EU regulations and directives that impose obligations on providers.
Multiple distinct contexts tend to collapse into one another or 'rot', degrading over time and reducing the utility of efforts to account for context.
Theoretical and empirical claim supported by interviewee reports and the authors' analytic synthesis; presented as observed pattern across cases (qualitative; sample size not specified).
Generative AI tools fail to account for users' context in workplace settings.
Findings from expert interviews reporting concrete examples where tools did not incorporate or respect relevant contextual information; qualitative analysis (sample size not provided in the summary).
Current approaches to account for the contexts in which generative AI technologies are used fall short of users' expectations and needs.
Qualitative empirical study based on expert interviews and analysis of user/developer perspectives (method described as expert interviews; exact sample size not stated in provided summary).
Occupations are not eradicated instantaneously, but gradually encroached upon via atomic actions.
Conceptual argument presented by the authors as part of their theoretical framing (Tech-Risk Dual-Factor Model); no empirical count reported for this specific claim.
Existing task-based evaluations predominantly measure theoretical "exposure" to AI capabilities, ignoring critical frictions of real-world commercial adoption: liability, compliance, and physical safety.
Authoritative statement in paper contrasting prior task-based exposure evaluations with the paper's focus on business/institutional frictions (liability, compliance, physical safety). No numeric sample; literature critique based on conceptual analysis.
Current research has largely focused on short-horizon tasks over a limited set of software with limited economic value (e.g., basic e-commerce and OS-configuration tasks).
Narrative literature/field observation reported in paper introduction (no numeric study reported in excerpt).
There is a fundamental gap in current agent capabilities: functional correctness alone is insufficient for design-aware issue resolution, motivating design-aware evaluation beyond functional correctness.
Synthesis of experimental findings: low design-satisfaction despite functional correctness, prevalence of design violations, and only partial improvement from guidance support the conclusion.
Design violations are widespread in agent-produced patches.
Empirical results from experiments on the benchmark showing many patches violate validated design constraints; backed by counts/percentages in evaluation (as summarized in abstract).
Test-based correctness substantially overestimates patch quality: fewer than half of resolved issues are fully design-satisfying.
Experimental evaluation with state-of-the-art LLM-based agents on the benchmark (reported in paper). Sample implicit: benchmark issues (495) used to evaluate agents; comparison between test pass rates and design-satisfaction measured by verifier.
Despite growing investment in data analytics, the decision-making and coordination layers of these workflows remain predominantly manual, reactive, and fragmented across outlets, distribution centers, and supplier networks.
Stated as an observation in the paper (abstract); no quantitative evidence, metrics, or comparative analysis provided in the excerpt.
Retail supply chain operations in supermarket chains involve continuous, high-volume manual workflows spanning demand forecasting, procurement, supplier coordination, and inventory replenishment.
Descriptive claim stated in the paper's introduction/abstract; no empirical data, sample, or methods reported to substantiate this characterization within the text provided.
The two margins interact through a self-undermining feedback that can generate low-archive traps (multiple equilibria with low accumulated public archive).
Dynamic equilibrium analysis in the theoretical model showing interacting feedbacks and possible trap equilibria (model-derived result).
Resolution margin: the probability that posted queries are resolved declines because AI raises contributors' outside options, thinning the contributor pool and creating congestion on the platform.
Mechanism and comparative-static implication produced by the paper's theoretical model; no empirical sample provided in the excerpt.
Flow margin: the posted volume of knowledge-enhancing queries declines as AI resolves more problems privately before they reach the platform.
Mechanism derived in the theoretical model; stated as the flow-margin channel (no empirical quantification in the provided text).
AI reduces archive creation through two distinct margins: a flow margin and a resolution margin.
Analytical decomposition derived within the paper's theoretical model (mechanism claimed by the model).
Generative AI resolves user problems without leaving a public trace, so fewer discussions and solutions reach public platforms.
Stated as an empirical motivation in the paper; no empirical sample or quantified measurement reported in the provided text.
The literature remains fragmented, with limited integrative frameworks to explain how AI-human dynamics and decision-making typologies shape outcomes.
Conclusion drawn from the systematic review and bibliometric analysis of the 627-article corpus as reported in the abstract.
Green AI research has largely measured the footprint of models rather than the downstream workflows in which GenAI is a tool.
Literature review / mapping of recent Green AI literature reported in the paper; descriptive claim about the focus of the field (no sample size or numerical counts reported in the abstract).
Existing benchmarks differ from real usage in programming language distribution, prompt style and codebase structure.
Paper asserts mismatch between existing benchmarks and production usage as motivation for producing a production-derived benchmark (stated differences: language distribution, prompt style, codebase structure).
Replacing deterministic components with probabilistic workflows changes the failure mode: LLM pipelines may generate plausible but incorrect outputs that pass superficial checks and propagate into irreversible actions such as DOI minting and public release.
Conceptual argument supported by the paper's incident descriptions (e.g., a detected coordinate transformation error); the statement is presented as a general risk rationale.
Up to 25% of routine administrative tasks face high automation risk.
Quantitative survey of 150 leading Nigerian firms across finance, tech, and manufacturing reporting the share of tasks at high automation risk.
There is a significant deficit in high-demand technical competencies such as data engineering, machine learning maintenance, and AI ethics within the Nigerian workforce.
Findings reported from the quantitative survey of 150 leading Nigerian firms (finance, tech, manufacturing) supplemented by qualitative workforce interviews and policy analysis.
The remaining 26 barriers are carried over from prior digital transformation waves — 22 in amplified form and 4 unchanged.
Comparative coding/classification within the review corpus indicating whether each barrier is novel or carried over, and whether it is amplified versus unchanged.
Three barriers were identified as agentic-specific: error propagation in multi-agent systems, role ambiguity, and accountability diffusion.
Classification of the 29 coded barriers by 'agentic specificity' within the literature review; these three barriers were labeled agentic-specific by the authors.
Occupations whose AI-exposed steps are more dispersed across the production workflow (higher fragmentation) exhibit a substantially lower share of their steps actually executed by AI, conditional on AI exposure share.
Empirical regression analysis controlling for share of AI-exposed steps; uses dataset linking O*NET tasks, human AI exposure assessments, Anthropic Economic Index execution outcomes, and GPT-generated workflow orderings (details in Sections 5.1 and 7).
In the limiting case of full automation, the model predicts that optimal recombination distance collapses to zero, suggesting that fully AI-driven research would undermine the very knowledge creation that it seeks to accelerate.
Limiting-case analytical result of the model: as the share of AI-automated tasks approaches 1 (full automation), the derived optimal recombination distance converges to zero.
Excessive reliance on AI may reduce the originality of research and lead to duplication of research efforts.
Model implication: as the share of tasks automated by AI increases, the paper shows analytically that originality can decline and firms may duplicate research efforts (due to homogenization of methods or search), reducing novel knowledge creation.
AI increases the aggregate rate of creative destruction, shortening the monopoly duration that rewards radical innovations.
Analytical result from the model: introducing AI raises the aggregate creative-destruction rate in the Schumpeterian framework, which reduces the expected monopoly duration and thus the rents that sustain radical innovation.
LLM uncertainty estimates require statistical correction before they can be used in decision-making.
Empirical finding of severe undercoverage of nominal 95% intervals and demonstration that conformal recalibration is needed to achieve intended coverage.
All models are severely overconfident: their 95% intervals contain the true value only 9--44% of the time, far below the expected 95%.
Analysis of model-produced 95% credible intervals across elicited population statistics, measuring empirical coverage rates reported between 9% and 44%.
There is a governance window—estimated at 10–15 years—before current deployment trajectories risk path-dependent social, economic, and institutional lock-in.
Forward-looking estimate/projection provided in the paper based on the authors' characterization of deployment trajectories and governance dynamics (no empirical sample size provided in the excerpt).
Societal consequences of labor displacement intensify the governance gap by concentrating consequential AI decision-making among an increasingly narrow class of technical and capital actors.
Analytic/theoretical claim in the paper drawing on the paper's multi-domain argument (no empirical sample size or quantified concentration metrics provided in the excerpt).
This nominal-vs-genuine oversight distinction represents the primary architectural failure mode in deployed AI governance.
Argumentative claim based on the paper's multi-domain synthesis and theoretical analysis; no empirical sample size or quantified causal inference provided in the excerpt.
The distinction between nominal and genuine human oversight is largely absent from current governance frameworks, including the EU AI Act and NIST AI Risk Management Framework 1.0.
Comparative policy/regulatory review claimed in the paper (explicit reference to the EU AI Act and NIST AI RMF 1.0); no sample size—based on textual/regulatory analysis rather than statistical data in the provided excerpt.
There exists a critical and underexamined governance gap between nominal human oversight of AI systems (humans in formal authority positions) and genuine human oversight (humans with cognitive access, technical capability, and institutional authority to understand, evaluate, and override AI outputs).
Conceptual/qualitative analysis and argumentation presented in the paper; implied synthesis of case examples and theoretical considerations rather than a quantified empirical study in the provided excerpt.
The accelerating displacement of human labor by artificial intelligence (AI) and robotic systems represents a structural transformation whose societal consequences extend far beyond conventional labor market analysis.
Stated as a framing claim in the paper; supported by the paper's literature review and multi-domain conceptual argument (no empirical sample size or quantitative data reported in the provided excerpt).
Sustaining such cooperative informational systems has historically proven difficult due to structural incentives that gradually erode transparency and trust.
Historical/analytical assertion in the paper; presented as a high-level observation (no dataset or empirical historical analysis provided in the excerpt).
The interaction between strict algorithmic control and worker counter-strategies leads to persistent limit cycles in strategy frequencies rather than convergence to a stable compliant workforce.
Dynamical systems analysis and simulation trajectories from the EGT model showing limit cycles / oscillatory equilibria in strategy proportions; model-based (no empirical sample).
The way we're thinking about generative AI right now is fundamentally individual (this appears in how users interact with models, how models are built, how they're benchmarked, and how commercial and research strategies using AI are defined).
Author's observational/descriptive claim supported by argumentative examples (mentions user interaction patterns, model design and benchmarking practices, and commercial/research strategies); no empirical sample or quantitative analysis reported in the excerpt.
Traditional questionnaires yielded slightly higher accuracy in risk assessment.
Result reported from the two experiments comparing traditional questionnaires to adaptive ARQuest versions; no numeric accuracy or sample size provided in the excerpt.
Insurers must blindly trust users' responses, increasing the chances of fraud.
Stated as a motivating problem in the paper; presented as logical/empirical concern rather than supported by a reported study within the paper.
Insurance application processes often rely on lengthy and standardized questionnaires that struggle to capture individual differences.
Descriptive claim in paper introduction arguing limitations of standard questionnaires; no experiment or sample size reported for this assertion.