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
Investments to build trust in AI (transparency, reliability, training) are likely to have positive returns via higher adoption rates and realized AI benefits.
This is presented as an implication derived from observed positive associations between trust and outcomes; the study did not conduct cost–benefit or longitudinal causal tests of such investments in the reported analyses.
Practical levers to increase AI trust include transparency of AI models, demonstrated reliability, and manager-focused AI literacy/training.
Paper proposes these levers based on study findings and discussion (recommendations), but they were not tested experimentally in the reported cross-sectional survey.
A stronger data-driven decision culture that stems from AI trust yields better operational and academic outcomes.
Study reports positive associations between AI trust → data-driven culture → operational and academic outcomes in survey-based analyses; however, the summary does not specify which operational/academic metrics were measured or sample size.
The dissertation implies policy interventions (subsidies, tax incentives, training and integration assistance) can accelerate welfare-improving AI adoption by helping firms overcome the early negative part of the U-shaped profit profile.
Policy implication derived from the theoretical U-shaped profit relationship and model interpretation; not supported by randomized or quasi-experimental policy evaluation in the provided summary.
Vendors that embed robust cognitive interlocks into development platforms can command premium pricing by reducing downstream risk; verification features may become a competitive moat.
Market-structure and product-differentiation reasoning in the paper; no market data, pricing studies, or competitive analyses presented.
Human verification (and automated verification infrastructure) becomes the limiting factor and a scarce complement to AI generation, raising demand and wages for verification expertise and tooling.
Theoretical labor-market analysis and complementarity argument in the paper; no labor market data or econometric estimates provided.
AI contributes to flatter, more networked and modular organizational forms, with increased cross-functional coordination enabled by shared data platforms and real-time analytics.
Conceptual reasoning supported by cross-sector illustrative examples; no standardized cross-firm comparative empirical study reported in the book.
Valuation of AI services should account for initiation assistance (fixed-cost reduction to starting tasks); monetizable value extends beyond direct task automation and could affect pricing/willingness-to-pay models.
Economic argument and implication drawn from the conceptual model; the paper does not provide empirical willingness-to-pay or pricing studies.
Conversational initiation assistance could complement human labor by increasing worker throughput and engagement, rather than directly substituting for skilled tasks.
Economic/managerial speculation in the paper; no empirical workforce or productivity studies presented.
Designing interfaces and metrics that focus only on task completion or execution misses value derived from initiation assistance.
Analytic recommendation based on the proposed model; no empirical metric-validation or A/B test results presented.
Conversational AI provides a distinct, non-executive mode of value — acting as an action-initiation interface in addition to being a task-execution tool.
Conceptual/economic argumentation in the paper; no empirical valuation or willingness-to-pay estimates provided.
Iterative conversation with AI surfaces sub-tasks and structures problems (structuring), creating clearer action plans and reducing initiation barriers.
Conceptual argument and illustrative example; paper does not present systematic coding, task analyses, or empirical tests.
Externalization (expressing frustration/stress to an external interlocutor) reduces affective load and decision paralysis, facilitating task start.
Theoretical reasoning supported by an illustrative anecdote; no empirical measurements or sample-based evidence provided.
Verbalization (talking through a problem with the AI) helps users organize thoughts and identify next steps, thereby lowering barriers to action.
Mechanistic argument in the paper; no experimental or observational data reported to validate the mechanism.
The 'Peripheral Approach' — beginning with casual, low-stakes dialogue (complaints, describing where one is stuck) rather than immediately requesting task execution — gradually reduces initiation friction.
Theoretical argument and illustrative anecdote from the author. No controlled studies or quantitative measures presented.
Casual, conversation-style interactions with AI can reduce psychological barriers that prevent people from starting tasks.
Conceptual/theoretical argumentation in the paper; illustrated by an anecdote (author's use of casual AI conversation to begin drafting the paper). No systematic empirical data, no experiments or observational samples reported.
Model and platform providers may capture significant rents through APIs and integrated developer tooling.
Market-structure analysis and observations of current platform monetization strategies; speculative projection based on platform economics.
Skill premiums may shift toward workers who can effectively collaborate with AI (prompting, verification, security auditing).
Theoretical and early observational studies suggesting complementary skills add value; limited empirical wage/earnings evidence to date.
Computer science curricula should emphasize computational thinking, debugging skills, and verification practices rather than rote coding alone.
Educational implications drawn from studies of learning with LLMs, risks of shallow learning, and expert recommendations; primarily normative and prescriptive rather than experimental proof.
When tasks are well matched to GenAI capabilities, firms can raise output per consultant and reduce time-per-task, thereby changing the marginal productivity of labor in consulting.
Inferred in the implications section from interview-based observations and the TGAIF framework; no reported quantitative measurement of output per consultant or time savings in the study.
Dynamic oversight regimes (ongoing audits, continuous certification) are likely more effective than one-time approvals for managing risks from agentic AI.
Policy and governance argument based on the dynamic nature of agentic systems; presented as a recommendation rather than empirically validated.
Firms will place greater value on alignment-as-a-service, monitoring platforms, and certification/assurance products as agentic systems proliferate.
Market-structure and demand reasoning from the paper; proposed as an implication rather than empirically demonstrated.
DAR-capable systems that credibly implement transparent registers and controlled reversibility may face lower adoption frictions in high-stakes sectors, affecting market dynamics and insurer/purchaser willingness to pay.
Economics-oriented implication and conjecture in the paper about adoption dynamics and market effects; not empirically tested in the manuscript.
Demand will increase for complementary goods: orchestration platforms, testing/verification tools, secure code-generation services, and team-level integrations.
Projected market implication based on practitioner-identified frictions (quality, security, integration) in the Netlight study; speculative market prediction without market data.
The need to orchestrate AI ensembles increases demand for skills in system design, AI-tooling, and coordination rather than only coding.
Authors' inference based on observed practitioner emphasis on supervision and integration tasks in the Netlight qualitative study; no labor market data provided.
First-mover and scale advantages are likely for firms that successfully integrate AI with robust oversight, potentially creating durable cost and service-quality advantages.
Theoretical and strategic analyses aggregated in the review; this is inferential and not supported by longitudinal competitive empirical studies within this paper.
Platforms combining high-volume generation with effective filtering/curation can create strong network effects and concentration in markets for AI-assisted ideation.
Market-structure reasoning and illustrative platform examples from the literature; no empirical market-wide causal studies reported in the review.
Firms that embed AI into collaborative workflows and invest in human curation may capture disproportionate returns (first-mover and scale advantages).
Theoretical/strategic argument supported by some applied case evidence and platform-market reasoning cited in the synthesis; the review notes absence of systematic causal firm-level evidence.
Generative AI will create complementarity: increasing returns to skills in evaluation, curation, synthesis, and domain expertise that integrate AI outputs.
Theoretical labor-economics reasoning supported by case studies and task-level studies showing demand for evaluation/curation skills in AI-assisted workflows; direct causal evidence on wage effects is limited in the reviewed literature.
Lowered cost and time of ideation and early-stage R&D due to generative AI may accelerate innovation cycles and reduce firms' search costs.
Inference from studies reporting reduced time-to-prototype and increased ideation; this is an economic interpretation rather than directly measured long-run firm-level innovation rates in the reviewed studies.
Firms must redesign KPIs to capture trust-related externalities (accuracy, escalation rates, repeat contacts) rather than only speed and throughput to avoid perverse incentives.
Recommendation based on observed trade-offs in deployments where emphasis on speed/throughput can harm quality/trust; not supported by randomized tests in the paper.
Transparency about AI use, seamless escalation to humans, and continuous monitoring/feedback loops are essential mitigations to avoid quality failures and trust erosion.
Governance literature, best-practice case studies, and deployment reports recommending transparency and escalation; limited direct causal evidence on mitigation effectiveness.
Firms that successfully integrate trustworthy, accurate AI can achieve faster strategic pivots and potentially gain competitive advantages and higher returns to organizational capital that embeds AI capabilities.
Associations between perceived trust/accuracy and organizational agility indicators in the quantitative analysis, plus qualitative case-like interview evidence suggesting competitive benefits; explicit causal estimates of returns not provided (implication is inferential).
Improved matching from predictive tools can shorten vacancy durations and improve reallocation dynamics in labor markets.
Implication from the review citing reported improvements in candidate screening and matching in some included studies; identified as a mechanism for labor-market effects.
k-QREM and its estimator provide useful behavioral primitives for applied AI-economics tasks (platform design, auctions, simulations), enabling richer modeling of boundedly rational agents and within-level heterogeneity.
Discussion and proposed applications section in the paper: authors illustrate potential uses and argue suitability based on the model's expressive structure and improved performance in numerical tests; no field experimental validation reported.
AI should serve precision and purpose in public policy — improving foresight, enabling better trade-offs, and preserving democratic accountability.
Normative policy prescription and conceptual argumentation in the book; no empirical testing or quantified outcomes reported.
AI-driven systems should empower people with knowledge and pathways to participate in global markets rather than concentrate gains.
Normative recommendation derived from policy analysis and value judgments in the book; not supported by empirical evidence in the blurb.
Authors propose the 'AI orchestra' concept: future development will involve coordinated ensembles of specialized AI agents (code generation, test generation, dependency analysis, security scanning) orchestrated by humans and higher-level controllers.
Theoretical/conceptual argument by the authors grounded in qualitative findings from Netlight (practitioner reports of multiple tools and coordination frictions); this is a forward-looking synthesis rather than an empirically established fact.
Canvas Design Principles aimed at reducing algorithmic myopia matter for welfare and regulatory concerns: better adaptive behavior reduces mispricing/misattribution risks but raises questions about transparency, accountability, and systemic amplification of shocks.
Policy and governance implication inferred from the claimed reductions in algorithmic myopia and increased adaptivity; study does not report direct welfare/regulatory impact measurements.
Faster, more accurate identification of demand shifts can compress the window for first‑mover advantages, intensify competitive dynamics, and raise the premium on organizational agility and human–AI integration capabilities.
Theoretical implication derived from observed improvements in signal detection (~5.8×) and resilience; not directly measured as market‑level competitive outcomes in the study.
Product teams evaluating LLM-powered features rely on a spectrum of practices—from informal “vibe checks” to organizational meta-work—to cope with LLMs’ unpredictability.
Qualitative interview study with 19 practitioners; thematic coding of transcripts produced descriptions of a range of evaluation practices used by teams.
Platform design choices (property rights, portability, reputation, tokenization, escrowed memories) will shape incentives for contributions to shared knowledge and agent improvement.
Policy and mechanism-design implications drawn from observed phenomena (shared memories, contributions, and trust) in the qualitative dataset; recommendation rather than empirically tested claim.
Shared memory architectures create public-good–like externalities (knowledge diffusion and spillovers) that may be underprovided absent coordination or platform governance.
Qualitative observations of shared memories and diffusion patterns plus theoretical economic interpretation; no empirical quantification of spillover magnitudes provided.
Adoption of C.A.P. may reduce demand for routine oversight/clarification roles and increase demand for higher-skill roles such as prompt/system designers and dialogue curators.
Labor demand and task composition analysis presented as a conceptual projection in the paper; no labor-market empirical study reported.
Because failure modes such as definition misalignment and hypothesis creep were observed, the authors argue for regulation/standards around disclosure of AI-assisted scientific claims and archival of verification artifacts.
Policy recommendation in the paper derived from the documented process-level failure modes in the single project; recommendation is prescriptive, not empirically validated beyond the project.
Lower data and compute requirements could decentralize innovation (reducing incumbent advantages tied to massive compute/data), but the complexity of embodied systems and real-world testing could create new specialized incumbents (robotics platforms, simulation providers).
Market-structure hypothesis based on trade-offs between resource needs and platform value; speculative and not empirically tested in the paper.
If smaller tuned models can capture most performance of much larger systems, market power may shift toward specialized, cheaper models plus toolchains, promoting niche competition and verticalized offerings.
Inference from empirical finding that a 7B tuned model achieves 91.2% of a larger model's quality; market-structure implication (theoretical/economic argument, not empirically tested).
There is a social welfare trade‑off between personalization value (higher AAR) and normative/social risk (higher MR); optimal policy and product design should balance these using BenchPreS metrics.
Analytical argument combining empirical findings (trade‑off between AAR and MR) with economic welfare considerations; the paper does not present formal welfare estimates or market experiments.
Organizational heterogeneity in strategic backing and mentoring explains variation in benefits from AI adoption across firms and sectors, contributing to cross-firm productivity dispersion.
Theoretical claim linking organizational moderators to heterogeneous adoption outcomes; proposed as an empirical research direction without data provided.
Managerial and peer mentoring styles (e.g., directive vs. developmental mentoring) influence how affordances are perceived and actualized, affecting learning, trust, and task allocation in human–AI collaboration.
Theoretical argument drawing on mentoring and organizational behavior literatures integrated with AST/AAT; no empirical tests or sample presented.