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
AI remains fragile for genuinely novel ideas, research-level experiments, and scientific judgment.
Summary claim from the paper's end-to-end lifecycle analysis indicating limitations on novelty and experimental rigor; no numeric performance metrics provided in excerpt.
Frontier LLMs fail to judge novelty reliably.
Paper's claim from its Validation-phase analysis that models do not reliably assess novelty; excerpt contains no underlying experimental sample or validation metrics.
Frontier LLMs miss hidden errors.
Qualitative statement from paper indicating models fail to detect some latent or subtle errors in research artifacts; no numeric evaluation provided in excerpt.
Under scientific pressure, even frontier LLMs still fabricate results.
Reported observation in paper about model behavior under scientific-use conditions; no specific quantitative experiments or sample sizes given in the excerpt.
Unrestricted frontier-scale checkpoint synthesis remains open (i.e., not yet solved).
Authors' assessment in the abstract noting current limits; asserts that unrestricted synthesis at frontier/model-scale has not been achieved.
AI agents deployed into SRE workflows currently derive their understanding of environment state from raw observability telemetry at query time, paying a semantic-interpretation tax in tokens, latency, and inferential reliability.
Author statement / problem framing in the paper (no quantitative experiment reported for this general claim).
Both major deployed systems and designed mechanisms concentrate on the most observable and easiest-to-govern tier, while the forms of commercial influence most consequential for user autonomy remain poorly understood and lack frameworks for detection, measurement, or disclosure.
Critical review of deployed system design choices and governance mechanisms; authors assert that attention and tooling focus on observable product-mention-level interventions while higher-tier influences lack measurement and disclosure frameworks.
These tiers instantiate across modalities and system architectures, including retrieval-augmented generation and agentic pipelines where upstream decisions can sharply constrain downstream outcomes.
Analytical claim supported by examples and discussion of system architectures (e.g., RAG, agentic pipelines) showing how interventions at different stages map to the taxonomy; no quantitative evaluation reported in excerpt.
Generative AI fundamentally changes advertising: rather than placing products into discrete slots, it enables interventions on the generative process itself, which induce commercial influence through less observable channels.
Conceptual argument backed by analysis of how generative models produce outputs and how interventions can operate on latent variables of generation; illustrated via taxonomy in the paper rather than quantified empirical tests.
Empirical research shows that ads woven directly into large language model (LLM) outputs often go undetected by users.
Reference to prior empirical studies (unspecified in the excerpt) showing user failure to detect embedded ads in LLM outputs; presented as an empirical finding rather than new experimental data in this paper.
Standard health system digital transformation policy, which typically addresses only the threshold failure through individual incentives, is predicted to systematically produce the partial adoption trap.
Model prediction contrasting full policy architecture vs. conventional policies that focus solely on individual incentives; analytical conclusion that such limited policies leave other failure modes unaddressed and therefore lead to stable partial adoption. Theoretical model; no empirical sample.
The barrier-lowering benefit of failed attempts is offset when trust erosion is rapid.
Model analysis combining cost-ratchet dynamics and trust erosion parameters; results showing interaction where fast trust erosion negates barrier reductions. Theoretical simulations/derivations; no empirical sample.
These failure modes are most severe precisely for the technologies with the greatest systemic value: the Value-Adoption Paradox.
Analytical result from the model showing failure-mode severity as a function of systemic value; theoretical identification of a paradox where higher systemic-value technologies face stronger coordination/trust/cultural barriers. Theoretical derivation; no empirical sample.
The basin of attraction of the partial adoption trap is enlarged by a cultural failure arising from negative coordination norms among doctors.
Model analysis including cultural coordination norms; theoretical demonstration that negative norms exacerbate partial adoption equilibria. Theoretical model; no empirical sample.
The basin of attraction of the partial adoption trap is enlarged by a trust failure arising from the organisation's inability to credibly commit to sharing productivity gains.
Model extension incorporating organisational commitment/transfer of gains; analytical results showing trust/commitment constraints increase stability of partial adoption. Theoretical model; no empirical sample.
The basin of attraction of the partial adoption trap is enlarged by a threshold coordination failure arising from the non-appropriable nature of systemic benefits.
Model analysis showing how non-appropriable systemic benefits (externalities) change payoff structure and enlarge the basin of attraction for partial adoption. Theoretical derivation; no empirical sample.
Current monolithic architectures struggle to enforce rigid brand constraints, frequently hallucinating unapproved visual assets.
Asserted critique of existing architectures in paper; no specific empirical metrics, datasets, or sample sizes provided.
Integration of generative video models into enterprise environments is restricted by temporal inconsistencies and severe brand misalignment.
Statement in paper describing deployment limitations; no empirical study, dataset, or sample size provided to quantify these restrictions.
Observed failures in the pilot were localized primarily to external integrations.
Pilot outcome summary in the paper stating failure localization was mainly due to external integrations (no numeric breakdown provided).
Agentic systems plan at inference time, making behavior sensitive to prompt variation and difficult to audit.
Author statement characterizing agentic (planning) AI systems and their inference-time sensitivity and auditability challenges.
Existing workflow platforms offer few semantic correctness guarantees.
Author statement contrasting current platforms' observability/durability with lack of semantic correctness guarantees.
Under an idealized model of independent steps, a ten-step process with 90% per-step reliability completes successfully only 35% of the time.
Analytic, idealized independence model reported in the paper (mathematical calculation: 0.9^10 ≈ 0.3487).
Selective displacement from AI is concentrated among older and lower-mobility workers.
Explicit claim in chapter summary, stated to be traced from labour market data and emerging workplace evidence (no numeric breakdown in excerpt).
Analysis indicates a significant negative relationship between perceived opportunities and challenges related to AI (i.e., higher perceived opportunities are associated with lower perceived challenges).
Correlation and regression analyses performed in SPSS on primary survey data showed a statistically significant negative association between measures of perceived opportunities and perceived challenges.
There exists employee resistance to change in response to AI adoption.
Survey-based measures of resistance included in the questionnaire and analyzed (descriptive/correlation/regression) using SPSS.
Employees identify ethical issues—particularly transparency and accountability of AI systems—as a notable challenge.
Survey items on ethical concerns analyzed with SPSS (descriptive and reliability analyses).
Employees have concerns regarding data privacy related to AI systems.
Primary survey data using a Likert-scale questionnaire; findings summarized with descriptive statistics and reliability analysis.
Employees report lack of AI-related skills (skill gaps) as a significant challenge to human–AI collaboration.
Survey responses from employees in AI-enabled organizations collected via a structured questionnaire and analyzed (descriptive/correlation).
Employees report fear of job displacement as a notable challenge associated with AI adoption.
Primary survey data (structured questionnaire) capturing perceived challenges; descriptive statistics reported.
The tech industry claims that its products, business models, and methods of resource extraction are unprecedented and fall outside any existing legal framework.
Descriptive claim about prevailing industry discourse referenced by the authors. (Citations or examples of industry statements not included in the excerpt.)
Exploitative working conditions violate workers' rights.
Legal assessment based on documents and the authors' interpretation of rights under applicable law (GDPR and labour rights frameworks). (Specific legal rulings or counts not provided in the excerpt.)
The results of this approach provide legally grounded evidence of the structural disadvantages faced by content moderators in the Global South, whose exploitative working conditions violate workers' rights.
Documents obtained via GDPR requests (employment contracts, NDAs, etc.) and legal interpretation are used as evidence to support claims of structural disadvantage and rights violations. (Specific documents and counts not provided in the excerpt.)
Current alignment approaches are primarily reactive rather than proactive.
Author's critique/characterization of prevailing alignment practice (conceptual observation without quantitative support).
The prevailing paradigm of alignment parallels early psychology's focus on mental illness: necessary but incomplete.
Analogy/argument presented by the authors as a conceptual critique (no empirical test reported).
Existing alignment research is dominated by concerns about safety and preventing harm: safeguards, controllability, and compliance.
Author's literature-level observation / conceptual review in the paper (no systematic review or quantitative coding reported).
Step-wise verification (verifying each stage of the reasoning chain) increases computational overhead and infrastructure requirements when deployed at scale.
Paper's structural trade-off analysis and engineering argument; no measured compute-costs, benchmarks, or sample-size reporting included in the provided text.
Process-based supervision introduces challenges regarding the sustainability of human-in-the-loop feedback loops.
Socio-technical argumentation in the paper—concern raised about ongoing human verification burden; no longitudinal or empirical data on human labor sustainability provided.
Deploying PRMs at scale introduces unique challenges regarding system latency.
Engineering and infrastructure trade-off analysis described in the paper; no measured latency benchmarks or sample-size performance tests provided in the supplied text.
Traditional outcome-based reward models, which evaluate only the final correctness of a solution, often fail to identify logical fallacies or "hallucinations" occurring within intermediate steps.
Theoretical critique and conceptual argumentation presented in the paper; no empirical study or sample size reported.
The findings carry direct implications for accountability, institutional integrity, and public trust in urban governance, and contribute to ongoing discourse on responsible AI adoption in cities aligned with global sustainability priorities.
Synthesis of audit results and discussion of their broader implications for public-sector adoption of LLMs in cities; inferential claim based on study outcomes (e.g., errors, fabricated sources, regulatory misinterpretation).
These failures extend beyond technical accuracy and introduce risks for governance, fiscal responsibility, and regulatory compliance.
Interpretation of audit findings (e.g., high rate of unverifiable citations, misinterpretation of regulations, degraded alignment on strategic scenarios) to argue systemic risks in governance and fiscal/regulatory domains.
Many responses misinterpreted regulatory requirements or relied on shallow justification.
Qualitative coding/analysis of LLM responses against expert rubric showing frequent misinterpretation of regulations and superficial reasoning.
Decision alignment with expert judgment degraded as scenario complexity increased, with strong agreement on operational triage but near-complete divergence on strategic capital allocation.
Comparative evaluation of LLM decisions vs. expert rubric across scenarios of varying complexity (operational triage through strategic capital allocation); qualitative and/or quantitative agreement measures reported in paper.
LLM self-reported confidence was negatively correlated with actual reasoning quality (r = -0.23), meaning the lowest-performing models projected the greatest certainty.
Statistical correlation reported between LLM self-reported confidence scores and measured reasoning quality across audited responses/models; correlation coefficient r = -0.23.
Across all models, 51.3% of cited sources were unverifiable or fabricated.
Quantitative audit of citations provided by the six commercial LLMs; proportion of cited sources judged unverifiable or fabricated as reported in paper.
No existing AI system replicates this: conversational recommenders treat recommendation as a terminal act, while general-purpose LLMs hallucinate product claims and default to generic promotional templates that fail to engage or persuade.
Author assertion/diagnosis comparing existing conversational recommenders and general-purpose LLMs; no empirical comparisons or quantified evaluation provided in the excerpt.
Accuracy is not a sufficient proxy for governance in regulated AI systems.
Empirical results from synthetic banking experiments showing divergence between task accuracy and governance-quality metrics across architectures, as summarized in the abstract.
Under text-only governance, 27% of deferrals carry no decision-relevant information.
Experimental evaluation in a synthetic banking domain comparing text-only governance to mechanical enforcement; reported statistic in paper abstract. Specific sample size not stated in abstract.
In algorithm-triggered emotional escalations, workers showed lower engagement: they sent fewer messages, contributed a smaller share of total chat rounds, and showed less proactivity in information seeking and solution provision.
Behavioral measures derived from chat logs in the randomized experiment comparing worker actions post-escalation across escalation types; reported differences in message counts, share of rounds, and proxies for proactivity.
Human intervention is less effective in algorithm-triggered emotional escalations (where customers express frustration or dissatisfaction).
Experimental subgroup analysis comparing intervention outcomes for algorithm-triggered emotional escalations versus technical escalations; emotional escalations showed worse post-intervention outcomes.