Evidence (16496 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
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 |
Organizations implementing AI without responsible transition mechanisms may worsen workforce anxiety, skill obsolescence, inequality, and trust erosion.
Paper's theoretical/conceptual assertion about risks of poorly-managed AI adoption; no empirical validation reported in the excerpt.
The International Monetary Fund estimates that nearly 40% of global employment is susceptible to AI, with exposure rising to 60% in advanced economies owing to cognitive task-oriented jobs.
Cited IMF estimate reported in the paper (reference to an IMF analysis; no sample size given in the excerpt).
Tenure negatively relates to AI use (OR = 0.846 per category).
Reported odds ratio from logistic regression for tenure categories predicting AI use; OR = 0.846 per tenure category.
The requirement that review + expected rework attention be lower than manual completion attention is substantially more stringent than the requirement that AI merely generate faster drafts.
Comparative analytical argument based on the model's derived stability conditions (theoretical/model-based reasoning; no empirical sample reported).
Under congestion, reviewers rationally raise the risk threshold for checking AI outputs, reducing scrutiny precisely when it would matter the most.
Analytical implication derived from the queueing model presented in the paper (theoretical/model-based inference; no empirical validation reported).
Mean-based metrics (e.g., tasks completed per worker-hour or mean handle time) can misrepresent AI's effects in workflows where tasks accumulate and compete for scarce human attention.
Argument and analysis presented in the paper; theoretical reasoning and illustrative queueing model (no empirical sample reported).
LLM-assisted discovery can increase report volume while maintainer-side validation, triage, funding, and release capacity may not scale—an effect that is acute in open source.
Claims supported by case material from Mozilla Firefox collaborations and Anthropic Mythos Preview public data, plus discussion of open-source maintainer constraints; no sample size given in the abstract.
The resulting bottleneck is not only finding more bugs; it is absorbing, validating, triaging, patching, and shipping a larger stream of reports.
Argument based on observed changes in report volume and workflow demands from public collaborations and market/program data referenced in the paper; exact empirical counts not provided in the abstract.
Regardless of apparent performance advances in AI technology, human and environmental factors of the organization may substantially attenuate — or even negate — the effective productivity benefits.
Conceptual argument in the paper; theoretical reasoning and literature synthesis (no primary empirical data reported in the abstract).
Adopting AI in organizational practice does not guarantee productivity gains, because human and environmental factors critically moderate the relationship between AI deployment and realized productivity improvements.
Position paper's conceptual argument presented in the abstract; no empirical sample or quantitative study reported.
Under water-constrained conditions, the framework achieves reductions of approximately 3-5% in generation-related freshwater withdrawals.
Quantitative results from simulation case studies on the IEEE test systems (reported percentage reduction ~3-5%); sample context: water-constrained simulation scenarios on IEEE 30-bus and 118-bus systems (sample_size = 2 test systems).
Because they are decoupled from the optimization process, static statistical accounting approaches are incapable of guiding workload relocation or power dispatch to mitigate water stress.
Argumentative claim in paper about limitations of static accounting methods with respect to guiding operational decisions (methodological critique).
Existing approaches typically rely on static statistical accounting to quantify these water footprints, but such static methods fail to capture how dispatch optimization and workload relocation dynamically affect water withdrawals.
Critical assessment in paper contrasting prior static statistical accounting approaches with dynamic needs; presented as methodological critique (no particular empirical sample in excerpt).
AI evaluation methods (benchmarks, red teaming, leaderboards) cannot be easily applied to human workers or yield comparable metrics.
Conceptual critique in the paper contrasting standard AI evaluation methods with human evaluation (no empirical comparisons provided).
Common criteria used to assess people (e.g., education, experience, references) cannot feasibly scale to AI systems.
Argumentative claim in the paper contrasting human hiring/evaluation practices with AI system assessment (conceptual; no empirical validation provided).
Human and machine workers may 'compete' for a given task, reproducing aspects of adversarial games.
Theoretical/assertional claim in the paper (conceptual discussion; no empirical data provided).
The increased use of algorithms in allocation decisions creates a Reverse Turing Test dynamic wherein the machine is now the judge.
Conceptual framing and argument presented in the paper (theoretical description; no empirical test reported).
AI-driven efficiency pressures in IT services may compress billable work and alter hiring and wage structures, raising transition risks even for technical workers.
Abstract cites high-reliability sector evidence (Reuters 2026a; Nasscom) to support this sector-specific claim; no sample size provided in abstract.
Labor-market segmentation and digital capability gaps in India create distributional vulnerabilities.
Abstract cites Indian official statistics and household/labor surveys (PLFS, HCES, MoSPI–NSO) and integrates sector evidence; no specific sample size reported in abstract.
Refined exposure measures imply widespread task transformation rather than uniform job destruction, with accelerated skill change as a central risk for vulnerable workers.
Abstract cites labor-market analyses and ILO (2025) as the basis for refined exposure measures and conclusions; no sample size stated in abstract.
Global frameworks warn that uneven readiness may produce a 'Next Great Divergence' between countries.
Cited global reports in abstract (UNDP 2025, WTO 2025, OECD 2026) which are summarized as issuing this warning; no primary data sample size reported in paper abstract.
Persistent adoption gaps among groups suggest unequal access to AI-enabled productivity.
Abstract references global reports (OECD, WEF, UNDP, WTO) and sector evidence indicating adoption gaps; no numerical sample size given.
AI may widen capability inequality—inequalities in access to knowledge, digital infrastructure, computational resources, and organizational adoption—thereby shaping income opportunities and socio-economic security for low-income groups.
Argument presented using the paper's socio-technical political economy framework and validated secondary sources (OECD, ILO, UNDP, WTO, WEF) and official Indian statistics; no direct empirical sample from this paper reported.
Design choices that prioritize scalable growth introduce trade-offs in reusability, evolution, and auditability in A2A collaboration networks.
Synthesis of empirical findings (low reuse, manipulable rankings, unverified validations) connecting design incentives to negative side-effects.
EvoMap relies on agents to provide local execution logs as evidence that uploaded assets function correctly; because these validations are not independently verified, over 84% of approved assets bypass quality checks using vacuous tests (e.g., console.log).
Empirical audit of validation logs and acceptance tests reported in the paper showing >84% of approved assets used trivial/vacuous checks.
Agents can trivially manipulate their asset's scores by falsifying self-reported metadata.
Demonstrations/analyses in the paper that changing metadata values leads to predictable changes in GDI scores; examples like claimed lines-of-code manipulation are provided.
An asset's GDI rank is heavily dictated by unverified, self-reported metadata (e.g., claimed lines of code modified).
Correlation/causal analysis in the paper showing strong dependence of GDI scores on self-reported metadata fields rather than objective performance measures.
EvoMap employs an algorithm (GDI) to score and rank shared assets, and this scoring system is flawed.
Paper description of the GDI ranking algorithm and empirical analyses illustrating problems with how it operates.
Rewards become highly concentrated among a small fraction of agents.
Distributional analysis of credits/rewards across agents (inequality/concentration observed in reward allocation).
98% of assets are never reused.
Empirical reuse metric computed across the asset corpus reported in the paper.
Because rewards favor publication over adoption, agents mass-produce assets to accumulate credits.
Observed publishing behavior (large numbers of assets per agent) and the platform's incentive structure; paper links publication-focused rewards to high per-agent asset counts.
Rewards are tied primarily to publication rather than adoption.
Analysis of reward allocation rules and empirical patterns showing reward issuance linked to publication events more than measured reuse/adoption.
These findings challenge the prevailing theory of skill-biased technological change.
Empirical observation that high-skill, high-exposure neighborhoods experienced wage stagnation post-2023 despite continued inflows of high-skilled workers, interpreted in contrast to predictions of skill-biased technological change.
Since 2023, high-exposure neighborhoods have experienced wage stagnation even as they continue to attract high-skilled workers (a 'high-skill trap').
Temporal analysis of job-posting wage signals in Beijing neighborhoods (2018--2024) using the GenAI Exposure Index to compare wage trajectories before and after 2023 between high- and low-exposure neighborhoods.
GenAI exposure is highly concentrated in the city's core districts, deepening the intra-urban AI divide.
Spatial analysis of a neighborhood-level GenAI Exposure Index constructed from 5 million Beijing job postings (2018--2024), where task-level assessments were aggregated across five leading large language models to measure exposure by neighborhood.
AI adoption contributes to labor market polarization and increases the risk of structural unemployment.
Authors' thematic synthesis of interdisciplinary studies reporting patterns of job polarization and macro/labor market risks associated with AI in manufacturing.
AI disproportionately affects routine and mid-skilled jobs.
Synthesis of literature (2010–2024) reported by the authors indicating disproportionate automation/AI exposure for routine and mid-skilled occupations.
AI adoption in manufacturing has critical implications for human labor, raising concerns about labor displacement.
Authors' systematic literature review (2010–2024) synthesizing interdisciplinary studies discussing labor impacts and displacement risks.
A-insensitivity acts as a cognitive barrier between beliefs and trust (i.e., it reduces the extent to which beliefs about forecast accuracy are translated into trust).
Interpretation based on experimental findings showing that higher a-insensitivity weakens the predictive relationship between beliefs about accuracy and expressed trust in analysts (derived from measures and analyses in the lab experiment; sample size not reported in abstract).
Decision-makers who are more a-insensitive are less likely to incorporate their beliefs about forecast accuracy into their trust judgments.
Experimental data where participants' a-insensitivity was measured and used to predict the extent to which their beliefs (optimism about accuracy) translate into trust for analysts (moderation/interaction analysis implied; sample size not reported in abstract).
AI adoption presents workforce adaptation challenges.
Reported in the study's literature synthesis and thematic analysis of secondary sources (qualitative review). No sample size reported.
AI adoption raises ethical considerations.
Authors' thematic evaluation of secondary literature identifying ethical issues associated with human-AI collaboration (qualitative synthesis). No sample size reported.
AI adoption presents challenges related to skill gaps.
Thematic findings from peer-reviewed literature and secondary data (qualitative review). No sample size reported.
Concentrated digital power may hinder inclusive industrialisation (SDG 9) and exacerbate global inequalities (SDG 10).
Argument linking conceptual analysis of digital power concentration to Sustainable Development Goals based on literature and policy interpretation (literature-based reasoning, no empirical measurement provided).
Industrial data systems generate 'participation without power,' a dynamic that particularly affects workers, small and medium enterprises (SMEs), and developing economies.
Theoretical/conceptual framing introduced by the paper and justified via literature review and examples from recent studies (no quantitative sample reported).
Inequality is increasingly shaped by the capacity to control and leverage digital systems rather than merely by access to digital technologies.
Conceptual claim grounded in synthesis of recent literature arguing a shift from access-based digital divide frameworks to control/power-based frameworks (literature review, no primary data reported).
There is a 'speedup illusion' where people have accurate forecasts of independent completion times but significantly underestimate AI-assisted times.
Empirical pattern reported in the abstract: comparison of predicted vs. actual times shows accurate independent forecasts but underestimation of AI-assisted completion times (preregistered study, N = 1237).
The fidelity gain from richer profiles comes with more input tokens per call from the longer prompts they require (i.e., higher per-call input cost).
Measurement of input token counts per model call for prompt variants with and without life-history profiles in the benchmark experiments; comparison shows longer prompts require more input tokens.
A conventional two-arm test understates the algorithmic channel by a factor of two.
Empirical comparison reported in the paper between the three-arm design estimates and conventional two-arm test estimates from the live campaign.
In the same campaign, the creative channel moves female impression share by -0.68 ppt.
Empirical result from the live Meta campaign reported in the paper; measured effect size (-0.68 percentage points).