Evidence (7278 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
9047 claims
Filter claims →
Productivity
8066 claims
Filter claims →
Governance
7278 claims
Filtered →
Human-AI Collaboration
6912 claims
Filter claims →
Org Design
4439 claims
Filter claims →
Innovation
4359 claims
Filter claims →
Labor Markets
3652 claims
Filter claims →
Skills & Training
3018 claims
Filter claims →
Inequality
2160 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 | 795 | 210 | 105 | 955 | 2131 |
| Governance & Regulation | 886 | 414 | 197 | 126 | 1654 |
| Organizational Efficiency | 826 | 204 | 129 | 87 | 1257 |
| Technology Adoption Rate | 681 | 259 | 128 | 110 | 1189 |
| Research Productivity | 464 | 138 | 65 | 349 | 1028 |
| Output Quality | 503 | 196 | 61 | 53 | 813 |
| Decision Quality | 351 | 180 | 84 | 51 | 673 |
| AI Safety & Ethics | 238 | 288 | 71 | 34 | 637 |
| Firm Productivity | 455 | 58 | 92 | 20 | 631 |
| Market Structure | 186 | 172 | 123 | 25 | 511 |
| Task Allocation | 222 | 70 | 76 | 34 | 407 |
| Innovation Output | 238 | 28 | 48 | 18 | 334 |
| Skill Acquisition | 177 | 62 | 62 | 17 | 318 |
| Employment Level | 107 | 57 | 108 | 13 | 287 |
| Fiscal & Macroeconomic | 135 | 72 | 44 | 26 | 284 |
| Firm Revenue | 172 | 50 | 28 | 5 | 256 |
| Consumer Welfare | 121 | 68 | 45 | 12 | 246 |
| Task Completion Time | 183 | 33 | 10 | 13 | 240 |
| Inequality Measures | 45 | 126 | 50 | 6 | 227 |
| Worker Satisfaction | 95 | 74 | 23 | 12 | 204 |
| Error Rate | 77 | 98 | 11 | 4 | 190 |
| Regulatory Compliance | 84 | 73 | 17 | 7 | 181 |
| Automation Exposure | 61 | 61 | 27 | 14 | 166 |
| Training Effectiveness | 98 | 21 | 14 | 19 | 154 |
| Wages & Compensation | 78 | 37 | 25 | 6 | 146 |
| Developer Productivity | 105 | 18 | 14 | 6 | 144 |
| Team Performance | 87 | 17 | 28 | 10 | 143 |
| Job Displacement | 12 | 83 | 23 | 1 | 119 |
| Hiring & Recruitment | 53 | 8 | 8 | 3 | 72 |
| Social Protection | 39 | 17 | 8 | 2 | 66 |
| Creative Output | 32 | 20 | 8 | 3 | 64 |
| Skill Obsolescence | 5 | 50 | 6 | 1 | 62 |
| Labor Share of Income | 17 | 20 | 17 | — | 54 |
| Worker Turnover | 15 | 15 | — | 3 | 33 |
| Industry | — | — | — | 1 | 1 |
Governance
Remove filter
Firms with a broader portfolio (greater diversity) of AI competitive actions achieve higher market valuation.
Empirical analysis using NLP-identified AI actions from S&P 500 firms' press releases (2010–2022); authors report positive association between portfolio breadth/diversity of AI actions and market valuation.
Firms engaging in more AI actions achieve higher market valuation.
Empirical analysis using NLP to identify AI competitive actions from press releases of S&P 500 firms over 2010–2022; authors report associations between the count of AI-related actions and firm market valuation.
Ultimately, this framework provides technology managers with a verifiable, evidence-based pathway toward resilient, net-zero Industry 5.0 ecosystems.
Conclusion/assertion in paper positioning the framework as a practical pathway; described qualitatively without empirical outcome measures or quantified evidence.
The architecture demonstrates how fabs can export cryptographically signed compliance tokens via International Data Spaces (IDS) connectors without exposing proprietary process recipes.
Claim of demonstration in paper; implies a prototype or illustrative workflow using cryptographic signing and IDS connectors, but no empirical deployment, sample, or measured disclosure-risk reduction reported.
By executing Virtual Metrology (VM) predictions and Federated Machine Learning (FML) inside hardware-rooted Trusted Execution Environments (TEEs), this architecture resolves the Data Sovereignty Paradox.
Technical claim based on proposed use of TEEs with VM and FML in the paper; presented as conceptual/architectural resolution rather than empirically validated result.
Structured as an interoperable network protocol stack, the framework coordinates an automated, five-step "relay race" between Facility, Process Engineering, and Finance proxy teams to align factory-floor yield models with macro-level sustainability mandates.
Architectural and protocol-level description in the paper (system design); no quantitative alignment metrics or empirical validation reported.
We propose a shift from reactive automation to autonomous governance through "Professional Proxies"—role-based agentic workflows executing within hardware-isolated trust zones.
Design proposal and conceptual workflow model presented in the paper; no field trial or user study reported.
We introduce a zero-trust socio-technical orchestration framework that operationalizes a six-layer SSbD reference architecture within trustworthy industrial data spaces.
Proposed system architecture described in the paper (design/proposal); no reported empirical deployment or quantitative evaluation.
Under managed growth pathways, infrastructure expansion could reach 226 GW.
Modelled managed-growth scenario(s) within the 21 scenarios reporting upper-bound additional capacity requirement.
Even under pessimistic scenarios, existing infrastructure would require 70 GW additional capacity.
Worst-case/pessimistic scenario results from the Europe-wide optimisation model (part of the 21 scenarios); reported additional capacity requirement.
In moderate scenarios, AI requires an additional 200 hours of firm generation.
Model scenario subgroup labelled 'moderate' within the 21 scenarios; reported additional firm generation hours requirement.
AI could drive 73-723 TWh of extra demand by 2050.
Spatially explicit optimisation model of Europe run across 21 AI growth scenarios; reported aggregate scenario range to 2050.
Scaling participation provides a technical foundation for transitioning from the monolithic status quo toward an open, bottom-up, and collaborative AI future.
Normative/architectural claim advanced in the paper arguing that the proposed paradigm enables a transition away from centralized monolithic models (framing/conclusion rather than an empirical result).
Participants contribute small models trained on their own interests and priorities; these models then collaborate in modular frameworks as compositional AI systems.
Description of the proposed 'scaling participation' paradigm and system architecture presented in the paper (design/method claim).
Participatory AI systems exhibit emergent capabilities that allow them to solve over 15% of problems where all individual models fail.
Reported experimental finding that the compositional system solved more than 15% of instances that none of the individual contributor models solved (exact problem count not provided in excerpt).
Participatory AI systems benefit from contributor diversity.
Further experiments reported in the paper that examine system performance as a function of contributor diversity (details not provided in the excerpt).
Participatory AI systems outperform monolithic LLMs by up to 15.4% across 15 tasks, such as reasoning and factuality, surpassing models larger than all contributed components combined.
Experimental evaluation reported in the paper comparing participatory (modular/compositional) systems to monolithic LLMs on a benchmark of 15 tasks (including reasoning and factuality); claims include quantitative comparison ("up to 15.4%") and a statement that the participatory systems outperform even models larger than the sum of contributed components.
Gains in institutional efficiency from GenAI adoption transmit to trading dynamics by stabilizing market volatility.
Empirical analysis in the panel linking institutional-efficiency improvements to lower/stabilized volatility metrics; methods include fixed effects, IV, and difference-in-differences.
Gains in institutional efficiency from GenAI adoption transmit to trading dynamics by increasing market liquidity.
Panel regressions linking estimated institutional-efficiency improvements to trading outcomes (liquidity measures) in the cross-market dataset; robustness checks with IV and difference-in-differences.
Institutional quality acts as a key moderating factor, amplifying the institutional-efficiency gains from GenAI adoption in markets with strong governance and regulatory capacity.
Interaction models and heterogeneity analysis in the cross-market panel; results reported for subsamples or interacted terms by governance/regulatory capacity measures.
GenAI adoption significantly improves institutional efficiency.
Cross-market panel dataset using a novel proxy for GenAI adoption; empirical approach includes fixed effects regressions, interaction models, instrumental variable estimation, and difference-in-differences designs to address endogeneity.
Shifting from post-hoc explanation to ante-hoc probabilistic mediation outlines a principled path toward AI systems that are not only powerful but fundamentally accountable.
Normative/conceptual claim about the implications and benefits of the proposed approach; no empirical validation presented.
Bayesian networks can encode domain knowledge, causal assumptions, and probabilistic dependencies before inference, enabling auditable reasoning traces, uncertainty quantification, and contestable outputs.
Theoretical claim about properties and benefits of using Bayesian networks as ante-hoc mediators; supported by conceptual argument and an illustrative benefit-eligibility scenario, not empirical measurements.
We propose the Glassbox Framework, in which Bayesian networks serve as transparent, ante-hoc mediation layers for generative models.
Design proposal/architectural contribution in the paper (conceptual framework specification); presented with a worked example but no empirical deployment or evaluation.
Large language models are rapidly becoming infrastructural components in high-stakes institutional settings, including public administration, legal reasoning, and healthcare.
Asserted as background/context in the paper (conceptual/observational claim); no empirical dataset or quantified adoption statistics reported.
The Recuse Signal, adapters, and experiment harness are released for reproduction.
Statement in the paper claiming release of the standard, adapters, and experiment harness (artifact availability claim).
In a controlled experiment pilot (SSH), the Recuse Signal cleanly induces recusal — 100% recusal when present versus 100% task completion in a no-signal control.
Controlled experiment / pilot described in the paper using SSH; deployed agents included OpenAI GPT-4o, GPT-4o-mini, and Anthropic Claude Code. Reported outcome contrasts between signal-present and no-signal control.
We implement two zero- or low-footprint adapters for the Recuse Signal: an SSH banner/PAM hook and a PostgreSQL wire-protocol proxy, and deploy them on a live production host.
Engineering implementation and live deployment described in the paper; two adapter implementations explicitly named (SSH banner/PAM hook and PostgreSQL proxy).
We propose a lightweight, published in-band deny signal — the Recuse Signal — that a server emits over a protocol's existing channels asking a connecting automated agent to voluntarily withdraw (a cooperative governance control, explicitly not a security boundary).
Design and specification proposed in the paper (open mini-standard); conceptual description and rationale; no prior empirical evidence required for the proposal itself.
Research should prioritise longitudinal and theory-informed evaluations, including intersectionality-informed analyses, and assess downstream impacts on women’s career trajectories alongside robust governance and accountability practices.
Authors' recommendations based on identified gaps from the scoping review.
Using inductive thematic analysis, we identified three functional domains: (1) bias mitigation and representation, (2) skills development and empowerment and (3) career pathways and retention.
Authors' thematic analysis of the 13 empirical studies included in the scoping review.
Artificial intelligence (AI) is increasingly integrated into career guidance and organisational decision systems.
Statement in abstract indicating observed trend; supported by literature search contextualising the review (scoping review using PRISMA-ScR).
Codeforces practice shifted toward this AI-style signature across cohorts over two AI rollouts.
Time-series/cohort analysis of CF practice data spanning two AI rollout periods (authors report cohort-level shifts; exact n not given in abstract).
Generative AI raises short-term productivity by completing tasks that learners would otherwise practice on their own.
Statement in paper's introduction/abstract; asserted as background premise (no specific sample size or empirical test reported in the abstract).
To foster more equitable outcomes, platform governance should be gender‑responsive, including algorithmic transparency, inclusive system design, and extension of core labor protections to gig workers.
Practical implications stated in the paper arising from the literature synthesis and feminist political economy framing.
AI‑enabled platforms can expand income opportunities and flexibility for women.
Thematic synthesis of findings across the 48 reviewed studies; reported in the paper's Findings as one side of a central paradox.
The proposed viewpoint reframes AI policy as the governance of an open, strategic, non-equilibrium learning system.
Conceptual reframing central to the paper; supported by the theoretical model and simulations presented.
The French AI debate should move beyond the binary opposition between techno-optimism and regulation-first caution.
Normative recommendation argued in the paper based on the HCLM framework and its implications for policy trade-offs.
The paper provides measurable policy indicators, game-theoretic propositions, illustrative simulations of national AI regimes, and concrete policy implications for France.
Claim about the paper's contents; the manuscript states that it includes indicators, propositions, simulations, and policy implications.
The paper connects HCLM with neural scaling laws, endogenous growth theory, creative destruction, and game theory.
Stated theoretical linkage and discussion sections; no empirical integration reported.
A competitive and human-centered AI strategy requires a controlled regime in which information injection grows faster than institutional dissipation while avoiding unstable, unequal, or energy-intensive expansion.
Prescriptive conclusion supported by the paper's mathematical model and illustrative simulations (model-based evidence rather than empirical causal estimation).
Information injection corresponds to compute, data, talent, research, capital, industrial deployment, and institutional experimentation.
Definition/mapping provided in the paper as part of the HCLM framework; conceptual rather than empirical.
AI sovereignty does not emerge from scale alone but from a country's capacity to regulate its own information dynamics.
Central theoretical claim supported by HCLM-based argumentation and linked conceptual arguments (neural scaling laws, endogenous growth theory, game theory); no empirical dataset reported.
France should be understood as a national AI learning system.
Conceptual/theoretical framing presented in the paper using Human-Centered Learning Mechanics (HCLM); no empirical sample or statistical test reported.
The paper concludes with policy recommendations to foster a conducive environment for AI integration, positioning Algeria to leverage technological advances for sustainable economic growth.
Concluding statement in the paper summarizing recommended policy actions; framed as guidance rather than empirically tested interventions.
Targeted investments and policy reforms could accelerate AI adoption and productivity gains in Algeria.
Policy recommendation inferred from the study's comparative findings and supported by citations to Brynjolfsson, Rock, and Syverson (2017) and McKinsey & Company (2023); presented as a prospective/conditional claim rather than an empirically estimated causal effect within the paper.
Artificial intelligence (AI) is rapidly transforming global economies by enhancing productivity, enabling innovation, and reshaping labor markets.
Framing claim supported by citations to Agrawal, Gans, & Goldfarb (2019) and Acemoglu & Restrepo (2020) as described in the paper's introduction; no primary empirical estimate reported in this paper.
This work highlights an urgent need for human-centric safety mechanisms that account for human factors, particularly in long-horizon, real-world development settings.
Authors' concluding claim based on empirical findings (high failure-to-detect, qualitative feedback) and design implications; normative recommendation.
Generative AI is being used for automation of tax compliance.
Listed in the abstract as an illustrative example of algorithmic application to international tax (generative AI for automating tax compliance); no empirical measurement reported in the abstract.
Blockchains are being used for instant trade verification in international tax contexts.
Presented in the abstract as one of three illustrative examples of how algorithmic technologies are being used for international tax purposes; no empirical details provided in the abstract.