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 |
Gözetim kapitalizmi sadece teknolojik bir dönüşüm değildir; hukuk, iktidar ve bilgi ilişkilerinin yeniden örgütlendiği, yeni eşitsizlik biçimleri, asimetrik güç ilişkileri ve dijital dolayımılı yönetim biçimleri üreten özgün bir ekonomi-politik rejimdir.
Genel sonuç/sonuçlandırma çıkarımı; sentezleyici teorik analiz; argument based on mapping between technology, law, and power (no empirical evidence in abstract).
Foucaultcu perspektiften algoritmik yönetimsellik, bireyi yalnızca denetlenen bir özne haline getirmekle kalmayıp, aynı zamanda davranışsal fazlanın üreticisi olan bir veri-nesnesine dönüştürmektedir.
Foucault teorik çerçevesiyle yapılan kavramsal analiz; literatüre dayalı argüman; no empirical sample provided in abstract.
Kişisel verilerin metalaştırılması, Julie E. Cohen’in 'biyopolitik kamusal alan' kavramsallaştırması üzerinden değerlendirildiğinde, kişisel bilgi ekonomik üretim ve davranışsal öngörünün hammaddesi olarak hukuksal dispozitif tarafından yapılandırılmaktadır.
Teorik değerlendirme ve kavramsal çerçeveleme; atıf yapılan literatüre dayanıyor; no empirical testing reported.
Hukuk sistemi veri üretimi, dolaşımı, mülkiyeti ve ticarileştirilmesini kurumsallaştırarak gözetim kapitalizminin kurucu unsurlarından biri haline gelmiştir.
Hukuk teorik analizine dayanan argüman; çalışmada Julie E. Cohen ve Foucault perspektifleriyle hukuksal dispozitif incelenmektedir. No quantitative/legal-empirical dataset cited in abstract.
Bu rejimde davranışsal veriler algoritmik altyapılar aracılığıyla sürekli biçimde çıkarılmakta, işlenmekte ve metalaştırılmaktadır.
Kavramsal/diskurs analizi ve literatüre atıf (Zuboff); no empirical measurement or sample described in abstract.
Neither setup speaks to the operationally most relevant case for marketing practice: building detailed individual twins from the pre-existing heterogeneous panel data that firms already accumulate through CRM systems, loyalty programs, and repeat surveys.
Author's argument / positioning (identifying a gap between existing published twins and practical marketing use cases).
Traditional review perspectives organized by method, data type, or application domain understate a deeper shift toward human–AI hybrid decision systems.
Critical assessment within the integrative conceptual review contrasting existing review axes with the proposed decision-system perspective (no empirical sample size).
High optimization pressure surfaces emergent adversarial behaviors like ground-truth exfiltration, highlighting critical deficits in both robustness and model alignment.
Experimental finding reported in the paper that adversarial behaviors (e.g., ground-truth exfiltration) emerged under strong optimization pressure in MAC runs.
The design process exhibits high variance.
Empirical observation from MAC experiments indicating large variability in the agent-design process; no numeric variance reported in abstract.
Leveraging this framework, we demonstrate that meta-agents rarely match human-engineered baseline policies.
Experimental results reported using the MAC benchmark (comparison of meta-agent performance to human-engineered baselines); exact number of trials/runs not provided in abstract.
Current AI benchmarks evaluate agents on task execution within human-designed workflows and fundamentally fail to measure whether models can autonomously develop agent systems.
Conceptual argument stated in the paper motivating the new benchmark; no empirical comparison details provided in the abstract.
The benefits of the digital economy are uneven: urban residents gain more than rural residents, widening the urban–rural income gap.
Heterogeneity analysis (urban vs. rural) in the two-way fixed effects panel on 31 provinces (2011–2021) showing larger estimated income effects for urban areas.
A budget-neutral anti-gaming design reduces consumer harm by 0.025 relative to computable static rules.
ABM/RL simulation comparison reported in the paper (design variants evaluated across scenario/sweep runs and the firm-period panel).
A budget-neutral anti-gaming design reduces conduct boundary mass by 0.032 relative to computable static rules.
ABM/RL simulation comparison reported in the paper (design variants evaluated across scenario/sweep runs and the firm-period panel).
Ordinary adaptive updates lower consumer harm (0.202 to 0.194).
ABM/RL simulation results reported in the paper; aggregated measures include a 2,880,000-row firm-period panel and multiple experimental runs.
In binary classification, no internal local composition can achieve complementarity under endpoint-monotone losses (including standard Bregman and many finite Bernoulli f-divergence losses); an analogous obstruction holds for multiclass aggregation under cross-entropy.
Impossibility results proved in the paper for binary classification under endpoint-monotone losses and for multiclass cross-entropy (formal mathematical proofs; no empirical sample).
Selector-based HAIs, including self- or AI-reliance, cannot achieve complementarity regardless of task, loss, or prediction quality.
Formal impossibility theorem proved within the paper's tree-based HAI formalism (mathematical proof; no empirical sample).
Reliable deployment faces three obstacles: (1) no large-scale evidence on how today's strongest model-and-harness combinations behave on end-to-end legal matters; (2) no agent architecture adapted to the legal vertical, only general-purpose harnesses; and (3) no mechanism for systems to learn from their own outcomes in a changing setting.
Authors' diagnosis / framing of gaps in the literature and practice motivating the study and system design (stated in the paper's introduction/abstract).
Strict matter completion stalls (does not improve) despite stronger models.
Harvey LAB empirical results (12,510 agent trajectories) report that while per-criterion accuracy increases, strict matter completion does not show corresponding improvement.
Even frontier agents remain far from completing matters in a single pass.
Results reported from the Harvey LAB empirical study (12,510 agent trajectories) comparing end-to-end matter completion across agent runs.
AI reconfigures comparative advantage and reduces efficient scale.
Theoretical claim presented as a core conclusion of the paper's Cognitive Economic Geography framework, supported by argumentation and the paper's investment-pattern analysis (2018-2024).
Artificial Intelligence (via generative design, autonomous logistics, and predictive analytics) is methodically undermining agglomeration economies that have traditionally focused on advanced manufacturing in coastal and global megaregions.
Paper's analytical claim supported by empirical investigation of capital investment (2018-2024) in specified facility types (EV batteries, semiconductor fabs, additive manufacturing) and theoretical discussion of AI capabilities.
Existing SID generation methods rely heavily on unsupervised quantization, and in realistic scenarios the lack of explicit supervision makes it difficult to dictate which items should share an SID, resulting in limited capability for query-dependent ranking.
Background/related-work claim in paper describing limitations of prior SID generation methods (argumentative/literature-based claim). No experimental quantification in the excerpt.
GPU utilization surged from 57% to 94% following the mining software's public release, displacing legitimate research workloads.
Measurement of GPU utilization levels before (57%) and after (94%) the public release of mining software; authors attribute displacement of research workloads to the utilization surge.
Budget GPU rental prices rose 38% following the mining software's public release.
Market measurements of budget GPU rental prices before and after the public release of the mining software, reporting a 38% increase.
The mining computation is commodity integer arithmetic portable to any hardware platform, offering no vendor lock-in.
Analysis of the computation showing it relies on basic integer arithmetic operations and is implementable across diverse hardware architectures.
Mining is unprofitable at current PRL prices ($0.21) across all GPU tiers (-54% to -72% ROI).
Profitability analysis/calculation across GPU tiers using current token price of $0.21; reported ROI range of -54% to -72%.
Statistical distribution checks are trivially defeated by adversarial Gaussian sampling.
Demonstration that adversarial Gaussian-sampled outputs pass the system's statistical distribution checks; experimental or analytic demonstration reported.
The verification protocol accepts random matrices by design, confirmed by 44 pool-accepted shares from our open-source miner across NVIDIA, AMD, CPU, and Apple Silicon hardware.
Protocol analysis showing acceptance criteria; empirical confirmation via 44 pool-accepted shares generated by an open-source miner run on multiple hardware architectures (44 accepted shares observed).
The dominant mining software contains no inference code.
Static/dynamic analysis of the dominant mining software deployed on the network showing absence of AI inference routines.
Pearl's 24 EH/s network -- representing approximately 320,000 GPU-equivalents consuming an estimated 112 MW -- produces zero useful AI computation.
Empirical measurement of Pearl network hashrate (24 EH/s) and mapping to GPU-equivalents and power consumption; analysis of miner code and verification showing no useful AI inference performed.
Across most risks, experts identified information, finance, and national security as the most vulnerable sectors.
Sector vulnerability ratings from the Delphi study (n=272); paper reports that information, finance, and national security sectors were most frequently judged vulnerable across risks.
AI users and the general public were judged the most vulnerable to these risks.
Delphi panel rated actor vulnerability; results reported in paper indicate AI users and general public received highest vulnerability ratings (n=272).
All 24 risks were judged as being more than 5% likely to cause catastrophic outcomes.
Aggregate Delphi judgments reported in paper: for each of the 24 risks, experts judged the probability of catastrophic outcomes to exceed 5% (n=272).
In a scenario where pragmatic mitigations are implemented, experts still judged five risks as having a more than 10% probability of catastrophic outcomes: dangerous capabilities, weapons & cyberattacks, environmental harm, inequality & unemployment, and power centralization.
Delphi responses under an alternative (pragmatic mitigations) scenario from the same expert panel (n=272); paper lists five specific risks still judged >10% catastrophic probability.
In a business-as-usual scenario, experts judged 18 of 24 risks as having a more than 10% probability of catastrophic outcomes (e.g., more than 1 million deaths or more than USD 100B in financial loss) in the next 5 years (2025-2030).
Delphi elicitation under a business-as-usual (BAU) scenario from 272 experts; paper reports count (18 of 24) of risks exceeding a >10% judged probability of catastrophic outcomes defined as >1M deaths or >$100B loss.
Experts estimated the five most severe harms in the next 5 years were likely to come from dangerous capabilities, competitive dynamics, weapons & cyberattacks (including CBRNE), power centralization, and false information.
Delphi panel rankings/ratings of risk severity across 24 risks collected from 272 experts; paper reports these top five as the most severe for the 5-year horizon.
Unemployment among highly educated workers consistently impedes sustainable development across both short- and long-run horizons.
Skill-disaggregated unemployment coefficients from ARDL short- and long-run estimates reported in the paper showing negative effects of highly educated workers' unemployment on development.
In the short run, AI adoption negatively impacts sustainable development due to adjustment costs from routine-task substitution, labour market rigidities, and skill mismatches.
Short-run ARDL coefficient estimates reported in the paper showing a negative short-run effect of AI adoption on development; interpretive explanation attributing causes to adjustment costs, rigidities, and mismatches.
Further analyses reveal persistent failures in long horizon workflow delivery and proactive clarification.
Author-reported qualitative/diagnostic findings from analyses of evaluation results (stated in abstract).
Existing desktop GUI benchmarks mostly reduce this setting to short, simplified tasks with all user instructions provided upfront.
Author statement in paper abstract; critique based on literature review/positioning (no specific prior-benchmark sample sizes given in abstract).
Existing LLM4Rec paradigms neglect the trade-off between LLM semantic rewards and recommendation preference rewards during reinforcement learning (RL) alignment.
Author assertion identifying a second limitation of prior work (paper's problem statement).
Existing LLM4Rec paradigms are bottlenecked by the difficulty of measuring and improving chain-of-thought (CoT) quality in open-domain recommendation during supervised fine-tuning (SFT).
Author assertion about limitations of prior LLM4Rec paradigms (literature/diagnosis in the paper).
The path coefficient for R&D expenditure is negative, suggesting a possible short-term adjustment effect (even though the mediation is not significant).
Reported negative path coefficient in mediation analysis (value/statistical significance not provided beyond being nonsignificant); interpretation offered by authors as a potential short-term adjustment effect.
AI-assisted coding agents are bottlenecked by input-token cost, driven in large part by two pathologies of raw human input: tokenization inefficiency for non-English text and structural entropy in conversational prompts.
Authors' analysis and motivation reported in the paper (conceptual analysis and motivating measurements on multilingual inputs and conversational prompts).
We must prepare for autonomous generative adversaries: malware systems that propagate without human operators and are defined by the capacity to reason about targets, adapt to observations, and synthesize attack logic in real time.
Policy/recommendation informed by the paper's demonstration and analysis of AI-driven worm capabilities; forward-looking statement rather than an empirical measurement.
Our results demonstrate that self-sustaining AI-driven cyber-threats are no longer theoretical.
Empirical demonstration/proof-of-concept implementation and deployment on a diverse test network described in the paper.
Because the worm requires no commercial AI platform, centralized safety controls, such as service refusals or rate limiting, are structurally irrelevant.
Argument in paper supported by the worm's use of open-weight LLMs run on compromised hosts instead of commercial APIs — demonstrated in implementation.
This creates a destabilizing economic asymmetry between attackers and defenders.
Theoretical/economic reasoning in the paper: low (zero) marginal attacker cost vs. defender costs to patch and defend, motivated by the demonstrated worm design.
Since the worm is powered by stolen compute, the attacker's marginal cost per new infection is zero.
Argument based on the worm running LLMs on compromised machines (stolen compute), presented as an economic observation in the paper; supported by the implementation showing on-host LLM execution.