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
Regulators and standard-setters who value transparency and auditability will need to account for the gap between evaluation results and actionable fixes; firms may require incentives or rules to ensure evaluation leads to remediation, not just documentation.
Authors' policy implication derived from the study's finding of a results-actionability gap and discussion of auditability concerns; speculative recommendation rather than empirical finding.
Delegation of oversight and reallocation of monitoring tasks due to AI integration changes transaction costs and affects organizational design and governance needs (e.g., more verification/audit effort or specialist oversight roles).
Based on participants' reported shifts in who performed monitoring/oversight tasks in the 40 interviews and the authors' interpretation of those shifts in organizational/economic terms.
The paper is the first systematic integration of XAI-based predictive modeling with counterfactual policy simulation specifically targeted at sustainability-oriented HR (Green HRM).
Authors' novelty claim stating this combination is novel in the Green HRM literature; no systematic literature review evidence provided in the summary to independently verify primacy.
The paper likely includes ablation studies and standard metrics (task success rate, step-wise error, plan coherence) to isolate contributions of the two training stages and to evaluate performance.
Summary states these analyses as 'likely additional methods' (i.e., typical but not fully detailed in the abstract); no direct confirmation or results provided in the provided text.
This study represents the first attempt to conduct a comprehensive evaluation of artificial intelligence (AI) and its influence on job displacement based on the existing body of literature.
Author assertion in the paper; the excerpt provides no external verification (no citation of prior reviews/meta-analyses to justify the 'first attempt' claim).
Results are robust across the authors' reported robustness checks.
Author statement that multiple robustness checks were performed and the main findings persist (the summary does not enumerate the checks or report their outcomes).
Observable firm-level and economy-wide moments—changes in spans of control, manager share of payroll, incidence of new tasks, employment growth, and shifts in the wage distribution—can be used to test the model's predictions.
Model-implied empirical identification strategy and suggested measurable moments in the paper's discussion/implications section (theoretical prediction, not an empirical test).
The PIER architecture (physics-informed state construction, demonstration-augmented offline data, decoupled post‑hoc safety shield) transfers to wildfire evacuation, aircraft trajectory optimization, and autonomous navigation in unmapped terrain.
Claim of transferability stated in the paper; the excerpt does not include experimental details or quantitative results for these domains.
Expect rising demand and wage premia for managers with hybrid capabilities (systems thinking + computational literacy), with a risk of widening returns to managerial skill heterogeneity.
Theoretical implication from predicted complementarities and task reallocation; prescriptive economic inference without empirical labor-market evidence in the paper.
Managers’ time will be reallocated toward hybrid tasks (interpretation, oversight, ethical deliberation), increasing returns to combined strategic and computational skills.
Predictive inference from the role reconfiguration analysis and task-complementarity argument; forward-looking theoretical forecast (no empirical time-use data).
Standards for provenance, labeling of AI-generated content, and interoperable evidence formats would lower verification costs and create beneficial network effects.
Policy recommendation derived from identified verification frictions and the study's analysis of data/model governance needs.
There is growing market demand for AI-assisted fact-checking tools, creating opportunities for software, monitoring services, and labeled datasets.
Analytic implication drawn from findings about increasing AI use and needs for automation/labeling; based on interviews and market inference in the study.
Regulators should consider guidelines on AI monitoring, algorithmic fairness in performance evaluations, and protections to prevent hybrid‑induced career penalties.
Policy recommendation based on conceptual assessment of risks identified in literature synthesis; not an empirical claim—no policy evaluation data provided.
Hybrid agency implies complementarity between GenAI and managerial/knowledge‑worker skills (curation, evaluation, coordination), potentially increasing returns to those skills while automating routine cognitive tasks—consistent with skill‑biased technological change.
Synthesis of recurring themes linking GenAI capabilities with managerial skill topics in the thematic clusters; positioned as an implication for labour demand and skill composition rather than an empirically tested effect.
There is demand for tooling that bridges evaluation outputs to actionable fixes (e.g., failure-mode libraries, standardized remediation templates, evaluation-to-priority mapping), signaling economic opportunities for third-party tools and consulting services.
Authors' inference based on the documented results-actionability gap and participants' descriptions of pain points; presented as a market implication rather than direct market measurement.
Firms that invest in instrumentation, cross-functional processes, and remediation levers capture more value from LLMs; organizations with better evaluation-to-action pipelines will obtain higher productivity gains and market edge.
Authors' inference from observed heterogeneity among teams in the interviews and comparison of practices in teams that reported more success converting evaluations into changes.
Public investments in standards, verification infrastructure, and public-interest datasets can correct market failures and support trustworthy AI.
Policy recommendation informed by governance and public-good theory and examples from the literature; the claim is prescriptive and not validated by new empirical evidence within the paper.
By lowering single-GPU resource requirements and improving throughput, SlideFormer can democratize domain adaptation and fine-tuning of large models on commodity single-GPU hardware (reducing the need for multi-GPU clusters).
Argumentative implication based on reported throughput, memory, and capacity improvements (e.g., enabling 123B+ models on a single RTX 4090 and reducing memory usage). This is an extrapolation from experimental results rather than a directly measured socio-economic outcome.
Regulators may prefer systems that support contestability and audit trails and could mandate argumentation-style explainability in certain sectors.
Speculative policy prediction; no regulatory statements or empirical policy adoption evidence cited.
Better contestability may reduce litigation and regulatory frictions if decisions are transparently defensible.
Speculative legal-economic claim; no case studies or empirical legal analysis provided.
New service layers may emerge (argumentation-as-a-service, audit firms, explanation certification, human-in-the-loop orchestration platforms).
Speculative market/industry evolution claim based on analogous tech-service cretions; no empirical evidence.
Collaborative VR features can change team workflows (remote, synchronous inspection sessions), potentially lowering coordination costs across geographically distributed teams.
Paper lists collaborative multi-user sessions as a planned capability and posits organizational effects; no user studies or measurements of coordination cost savings presented.
Public funding for shared VR-capable data-exploration infrastructure could yield high leverage by improving returns on large observational investments.
Policy recommendation deriving from the platform and ROI arguments in the paper; no cost-benefit analysis or quantified ROI provided.
Using iDaVIE increases the usable fraction of large observational datasets by improving QC and annotation throughput, thereby raising returns to telescope investments and downstream AI efforts.
This is an inferred implication in the paper (returns-to-scale/platform effects) based on improved QC/annotation throughput; no empirical measurement of usable-fraction increases provided.
Higher-quality labels produced via immersive inspection can reduce label noise and lower required training-data sizes for a target ML performance level.
Paper presents this as an implication/expected outcome based on improved annotation quality from immersive inspection; no empirical ML training experiments or quantitative reductions reported.
iDaVIE demonstrably reduces cognitive load for multidimensional-data tasks compared with 2D-slice inspection.
Paper asserts reduced cognitive load and faster, more intuitive exploration as an aim and reported outcome; no formal user-study metrics, sample size, or statistical analysis provided.
The inverse-specification reward offers a domain-agnostic, holistic metric for fidelity to user intent and is recommended for measurement of model value/service quality.
Method introduces inverse-specification reward and asserts domain-agnostic applicability; recommendation based on its conceptual ability to recover briefs as fidelity measure (not necessarily validated across many domains).
High-quality automated slide generation has potential to reduce time spent on business presentation creation and produce productivity gains with partial substitution of routine creative/knowledge-worker tasks.
Empirical demonstration of near-SOTA automated slide generation capability on 48 briefs; domain-level economic implication extrapolated from performance improvements.
Deploying BATQuant with reliable 4-bit weight/activation quantization for MXFP-capable accelerators reduces memory footprint and memory-bandwidth pressure, enabling higher throughput and lower per-token inference costs.
Argumentative / economic analysis in the paper linking reduced precision and parameter storage to lower memory/bandwidth requirements and inferred throughput/cost improvements; not presented as a direct empirical measurement of cost per token in production environments in the summary.
The methodological template (train an ML surrogate of a costly simulator and embed it in an optimizer) generalizes beyond Doherty power amplifiers to other analog/microwave components and broader engineering domains.
Paper proposes generality of approach in implications section; no experimental demonstrations beyond the Doherty PA case are provided in the summary.
Investment in data quality and feature engineering yields tangible predictive gains for workforce performance models.
Paper emphasizes use of engineered features capturing engagement dynamics and learning trends and reports better model performance relative to baseline; however, no isolated ablation study quantifying the sole contribution of data-quality investments is reported in the summary.
Tools that improve detection or quantification may reduce downstream costs from missed diagnoses or unnecessary follow-ups, improving cost-effectiveness in some scenarios.
Economic modeling and limited observational analyses that extrapolate diagnostic improvements to downstream resource use; direct empirical cost-effectiveness studies are scarce.
The metacognitive reliability metric can reduce adoption risk for purchasers by providing transparent error-risk assessments and enabling performance-based autonomy thresholds.
Conceptual claim supported by the existence of an empirical confidence metric from the recursive meta-model and discussion of procurement/decision-making implications; not empirically tested with purchasers or procurement outcomes.
HACL/CS supports human trust and situational awareness.
Human factors measured with trust and situational awareness questionnaires in the simulation; summary reports supportive effects on trust and situational awareness but lacks sample-size/statistical detail.
Intelligent turn-level assignment can reduce costly human attention to only high-value moments, improving overall system productivity.
Conceptual implication from the assignment-layer design and empirical trade-offs reported; presented as an advantage in the paper rather than a directly measured economic productivity study.
HADT demonstrates a concrete way to substitute expensive human diagnostic labor with AI assistance while preserving high accuracy, implying reductions in marginal cost per consultation.
Inference drawn in the paper's implications section based on reported reductions in required human effort and maintained diagnostic accuracy (economic claim extrapolating from experimental results; not directly measured as cost in experiments).
Organizational norms and UX influence adoption rates and diffusion of AI: social calibration processes at the team level matter for adoption beyond individual cost–benefit calculations.
Reported by interviewees (N=40) as factors shaping whether and how teams incorporated AI into routines; integrated into theoretical implications for diffusion modeling.
Well-calibrated trust tends to encourage AI being used as a complement to human labor (augmentation), increasing effective productivity; miscalibration (over- or under-trust) can lead to productivity losses.
Inferential claim drawn from interviewees' accounts of when teams appropriately relied on AI (augmentation) versus when inappropriate reliance or avoidance occurred; supported by thematic interpretation rather than quantitative measurement.
Policymakers should support standards for auditability, human‑in‑the‑loop thresholds and training subsidies to reduce coordination failures and make the social benefits of AI adoption more widely shared.
Normative policy recommendation derived from the paper’s analysis of risks, governance needs and distributional concerns; not empirically validated within the paper.
Organisations will invest more in training for AI‑related sensemaking, trust calibration and governance competencies; returns to such training should be evaluated relative to investments in model quality.
Prescriptive inference from the framework and human‑capital theory; supported by referenced literature but not empirically tested in this paper.
Explicit comparative‑advantage allocation will shift the composition of tasks across humans and AI, altering demand for routine versus non‑routine skills and potentially increasing demand for high‑level judgement, oversight and sensemaking skills.
Projected labour‑market implication based on theoretical reasoning and prior literature on task‑based skill demand; not empirically estimated in the paper.
Operationalising the four symbiarchic practices through updated HR systems lets firms capture AI‑enabled productivity gains without eroding trust, ethics or employee well‑being.
Normative claim based on theoretical synthesis and managerial prescription; no empirical testing or field data presented in the paper.
Public data sharing, reproducibility standards, and shared benchmarks could raise the floor of AI utility across the industry.
Policy implication grounded in arguments about data quality, coverage, and generalizability from the narrative review; speculative recommendation rather than evidence-backed empirical claim.
There is potential for consolidation as firms acquire data, talent, or validated AI-driven assets.
Industry-structure implication drawn from economics of complementary assets and observed M&A activity patterns; presented as a likely trend rather than demonstrated empirically in the paper.
AI startups that demonstrate validated, reproducible wet-lab outcomes and access to high-quality data are more likely to command premium valuations.
Argument from observed market behavior and economics of complementary assets presented in the narrative; no systematic valuation analysis included.
Investors should recalibrate expectations: greater value accrues to firms that integrate AI with experimental pipelines and proprietary data assets rather than firms that only possess AI capability.
Economics-focused implications drawn from thematic analysis of heterogeneity in firm outcomes and integration requirements; market-practice inference rather than empirical valuation study.
By integrating psychological trust factors with cognitive capability optimisation, this model offers actionable insights for knowledge management practitioners implementing AI‑augmented decision systems while advancing theoretical understanding of human–AI collaboration effectiveness.
Integrative theoretical claim based on combining constructs from psychological trust research and cognitive/capability literature via systematic synthesis; no empirical evaluation reported in the abstract.
The framework provides practical guidance for executives designing human–AI teams, developing trust calibration training, and establishing performance metrics.
Prescriptive recommendations derived from the proposed model and literature synthesis; the abstract does not report empirical testing of the recommended interventions or their effects.
Supportive regulatory frameworks and digital infrastructure development are important for leveraging AI technologies to improve global trade efficiency.
Study recommendation derived from empirical findings and discussion; this is a policy implication rather than a directly tested empirical claim (no policy evaluation data provided in the summary).
The study provides empirical support for digital transformation theories within financial intermediation.
Authors interpret quantitative results as empirical evidence consistent with digital transformation theories; specific theoretical tests, model fit statistics, and sample information are not included in the summary.