Evidence (8807 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
Filtered →
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 |
Productivity
Remove filter
Pooled across five AI coding agents, pull requests (PRs) with a human Co-Authored-By trailer merge less often than purely-autonomous ones (53.8% vs. 79.8%).
Aggregate analysis of PR merge rates across five AI coding agents in the AIDev dataset; pooled sample of PRs (33,596 PRs referenced elsewhere in the paragraph).
The paper identifies two distinct gaps that have widened as GPTs exposure scores traveled from their time and place of production: (1) a structural gap between what static exposure scores measure and what policy questions require, and (2) a coordination gap between researchers and policymakers.
Explicit framing and thesis presented in the paper summarizing the central arguments.
Policy-relevant work that asks who is harmed or benefits, how, and when continues to reference static GPTs exposure scores without engaging with methodological updates needed to answer these questions more reliably.
Critical literature review and observed citation practices reported by the authors; claim based on review of how policy analyses cite/ use the scores.
These temporal, geographic, and ontological limitations compound when exposure scores are used in policy-facing analyses.
Conceptual argument and case-study approach in the paper showing how limitations interact and worsen policy analysis outcomes.
The GPTs exposure scores have temporal, geographic, and ontological limitations that do not always travel with the scores as they are reused.
Authors' methodological critique discussing the limits named by Eloundou et al. (2023) and how those limits are often ignored when scores are repurposed.
Critical post-work thought posits that not only certain jobs, but also jobs in general, are disappearing.
Statement summarizing the position of a body of theoretical work ('critical post-work thought') as described by the author; this is a characterization of a viewpoint rather than an empirical finding.
The majority of extant studies focus exclusively on the 'technical' aspect of new technologies replacing labour, thereby ignoring their social dimension and consequently falling into the trap of technology fetishism.
Claim about the literature based on the paper's review and critique of existing studies; no citation counts or systematic review methodology described in the excerpt.
Existing frameworks address AI-assisted development maturity or the productivity-reliability tension but offer no mechanism for calibrating human oversight intensity to regulatory impact.
Comparative framework analysis and literature review reported in the paper (claims about gaps in existing frameworks).
The adoption of agentic AI coding systems -- where autonomous agents generate, review, test, and deploy code with minimal human intervention -- creates a governance challenge in regulated industries.
Argumentation in the paper framing the problem; conceptual analysis of agentic AI capabilities and regulatory constraints (literature/contextual reasoning rather than empirical data).
Policy asymmetries, digital literacy gaps, and regional inequalities deepen digital divides and impede inclusive development.
Policy analysis and comparative case studies documenting how policy differences, literacy, and regional disparities affect digital inclusion; China used as a focal example. No quantitative sample sizes or causal estimates given in summary.
Agriculture remains digitally marginalized due to infrastructural and institutional deficits.
Comparative case studies and sectoral data showing lower digital adoption in agriculture; qualitative policy analysis identifies infrastructure and institutional shortcomings. No sample size or quantified adoption metrics provided in summary.
Fertility is strongly countercyclical and almost perfectly negatively correlated with hours worked in the model, placing household time allocation at the center of the mechanism.
Model-simulated correlations and business-cycle dynamics showing fertility and hours worked time series and their correlation.
The longevity shock compresses asset returns and lowers the real interest rate, and generates hump-shaped, persistent dynamics.
Numerical impulse-response dynamics from the overlapping-generations model following a longevity shock; reported time paths for returns and the real interest rate.
Manual preparation of engineering designs for thousands of wells constitutes an enormous administrative burden and is prone to inconsistencies.
Introductory/background statement in the paper describing the pre-existing manual workflow burden; no numerical study reported for this specific statement.
A 'critical transmission path' can occur in which AI-induced productivity gains are weakly transmitted to households and may generate absorption tension.
Conceptual framework / theoretical argument in the review (no empirical sample reported).
Productivity gains from AI do not automatically translate into broadly distributed welfare or into output fully absorbed by market demand.
Conceptual review / theoretical argument and literature synthesis presented in the paper (no empirical sample reported).
Adding relevant collaborators can lower performance when teams lack structure to coordinate their contributions.
Empirical comparisons across experimental sessions in the Collaborative Gym / DiscoveryBench setup; result reported across the study (1,482 sessions).
A wide range of empirical evidence shows that humans avoid complexity, delegate judgement, and prefer simplified social worlds.
Asserted as empirical background; paper references a broad empirical literature but does not report primary data, sample sizes, or specific studies in the provided text.
Mainstream multi-agent hierarchical decision architectures often rely on coarse-grained instructions that underspecify analytical procedures, leading to degraded inference quality and reduced transparency.
Framing/critique stated in the paper about prior approaches; no empirical comparison statistics provided in the excerpt to quantify the extent of degradation.
Most organizations (59%) approach AI implementation through a technology-first lens, layering intelligent systems onto legacy processes rather than intentionally redesigning how humans and machines collaborate.
Reported descriptive statistic from Deloitte's 2026 Global Human Capital Trends survey of over 3,000 business leaders across 15 countries (paper cites 59% figure).
Only 14% of organizational leaders report proficiency in designing effective human-machine interactions.
Reported descriptive statistic from the same Deloitte 2026 Global Human Capital Trends survey of over 3,000 business leaders across 15 countries.
Current machine learning models commonly require large and well-annotated datasets, and the annotation process often becomes a bottleneck with increased complexity leading to higher chances of human errors.
Background statement in the paper summarizing common knowledge and prior literature about dataset requirements and annotation challenges.
Expertise moderated the effect of LLM guidance: novices exhibited passive AI reliance.
Stratified analyses by participant expertise level using behavioral and eye-tracking measures indicating novices shifted attention to the AI/chat and exhibited more passive acceptance of guidance.
AI augmentation breaks the accounting link between labor time and productive contribution, yet firms continue to evaluate talent through time-based overhead bundles.
Theoretical argument and conceptual framing presented in the paper (no empirical sample reported for this specific proposition).
Investment is being directed toward AI deployment when achieving productivity gains requires prior development of convergence capacity (C), leading to a misallocation of investment.
Theoretical reasoning within the paper: conceptual argument that deployment-focused spending misses prerequisite cognitive capacity (C).
Prevailing production-function frameworks encounter a structural boundary because they treat AI as a separable factor of production without modeling the cognitive mediation through which AI generates productive value.
Theoretical / conceptual argument presented in the paper (derivation and critique of existing production-function approaches).
Massive AI investment has failed to generate commensurate productivity gains (the "AI productivity paradox").
Stated as the motivating empirical paradox in the paper; presented as an observed phenomenon motivating the theoretical argument (no specific dataset or numeric evidence provided in the abstract).
The inhibitory effect of computing power deployment on corporate financialization spills over from the host city to surrounding cities.
Spatial Durbin model estimation showing negative effects on financialization in neighboring cities around NSC locations.
The reduction in corporate financialization following computing power deployment is concentrated in speculative financial assets rather than precautionary financial assets.
Subsample/asset-type analysis distinguishing speculative vs. precautionary financial asset holdings in firm balance sheets and estimating differential effects.
The inhibitory effect on financialization is more pronounced for firms with low analyst coverage.
Heterogeneity analysis splitting sample by analyst coverage and estimating differential DiD effects.
The inhibitory effect of computing power deployment on corporate financialization is stronger in computing-intensive industries.
Heterogeneity analysis comparing treatment effects across industries with different computing intensity using the staggered DiD setup.
Computing power deployment significantly reduces corporate financialization levels by approximately 1.1 percentage points.
Empirical analysis on Chinese A-share listed companies (2012–2023) exploiting staggered establishment of National Supercomputing Centers (NSCs) as quasi-natural experiments and estimated using a staggered difference-in-differences model.
The translation of AI's potential into operational capability within government audit contexts requires navigating complex technical, institutional, legal, and ethical challenges that differ substantially from private sector environments.
Paper's conceptual analysis and comparative argument (paper contrasts government audit contexts with private sector origins of many AI tools); no quantitative empirical evidence or sample size reported.
Scaling per-user LLM profiling to a live, millisecond-latency dispatcher faces three constraints: logs exceed any LLM's context window by orders of magnitude; most users are long-tail, with too few interactions for per-user profiling; and surface-fluent profiles do not necessarily improve downstream prediction utility.
Problem motivation and observational claims stated in the paper describing practical constraints; empirical quantification of these constraints is not provided in the abstract.
These growing interconnections create new vulnerabilities that can spread across public service networks.
Systems-theory informed synthesis from the review of empirical literature; paper's integrative conceptual framework drawing on reviewed studies.
The canonical manifestation of this failure pattern is called 'Phantom Legislation' (internally consistent symbolic outputs disconnected from real business semantics).
Terminology and descriptive example provided by the authors based on their analysis of observed failure cases in the Bang-v3 project.
GPU-accelerated deep learning exacerbates this problem, as nondeterministic floating-point reductions can produce drift in long backtests, challenging regulatory reproducibility and auditability expectations.
Argument in paper linking known GPU-nondeterminism (floating-point reduction nondeterminism) to practical issues in long financial backtests; no empirical backtest dataset size provided in the excerpt.
For thirty years, quantitative finance has paid a costly two-language tax: models researched in Python are rewritten in C++ for production, often introducing numerical discrepancies.
Statement in paper's introduction/abstract describing historical practice; no quantitative sample size or systematic study reported in the excerpt.
Test file counts substantially overestimate verification strength.
Conclusion drawn from the high prevalence (80.2%) of test patches with weak/no oracle signals compared to mere presence-of-test-file counts.
Raw merge rates are lower for strong-oracle PRs.
Unadjusted (raw) comparisons of merge rates between PRs classified by oracle strength in the study dataset.
Applied at scale, 80.2% of test patches contain weak or no explicit oracle signals.
Automated/syntactic classification of oracle-signal categories applied to the full dataset of test-file patches (as described in methods).
Test files lacking explicit assertions execute code without verifying behavior, so quality gates based on test-file presence overestimate verification strength.
Conceptual/analytic argument supported by the paper's framing and subsequent empirical analysis of oracle signals.
Optimization variance admits a necessary lower bound on its decay rate, implying fundamental constraints on how quickly uncertainty dissipates regardless of the predictor used.
Mathematical proof/theoretical result presented in the paper (formal lower-bound derivation; no empirical sample size in excerpt).
The high cost of fine-tuning LLMs poses a significant economic barrier.
Stated in the paper's motivation/intro as a high-level observational claim (no sample size or numeric cost figures provided in the text excerpt).
The AI-investment paradox persists because firms govern AI as a broad technology program rather than as a set of discrete, investable decision opportunities embedded within workflows.
Argument/theoretical claim developed by the authors as the central explanatory hypothesis of the paper; presented conceptually rather than tested empirically within this work.
Despite enterprises continuing to invest heavily in AI, many initiatives fail to scale or generate sustained business value (the 'AI-investment paradox').
Background claim stated in the paper's introduction/abstract and presented as motivating fact; supported implicitly by citations to prior literature and industry reports (no original empirical sample or quantitative analysis reported in this paper).
Developing a process twin is costly because it requires accurately modelling the entire production process (process steps, equipment and product-specific settings, process variations) and binding the model to live operational data.
Authoritative/technical description of development requirements and costs in paper (methodological claim), not quantified in abstract.
In a two-type heterogeneous-agent economy, high-cognitive-capital agents adopt AI more intensively and may eventually erode their unaided cognitive capital below that of initially lower-skilled agents.
Heterogeneous-agent extension of the analytical model; stated as a derived proposition. No empirical validation.
The decentralised equilibrium over-adopts substitutive AI relative to the social optimum because of systemic risk, cognitive public goods, and arms-race externalities.
Equilibrium analysis and welfare comparison in the formal model (decentralised vs socially optimal allocation). No empirical sample.
Expected crisis losses are convex in aggregate leverage.
Analytical result/proposition derived within the model showing convex relationship between expected losses and aggregate leverage. No empirical sample.