Evidence (3103 claims)
Adoption
5267 claims
Productivity
4560 claims
Governance
4137 claims
Human-AI Collaboration
3103 claims
Labor Markets
2506 claims
Innovation
2354 claims
Org Design
2340 claims
Skills & Training
1945 claims
Inequality
1322 claims
Evidence Matrix
Claim counts by outcome category and direction of finding.
| Outcome | Positive | Negative | Mixed | Null | Total |
|---|---|---|---|---|---|
| Other | 378 | 106 | 59 | 455 | 1007 |
| Governance & Regulation | 379 | 176 | 116 | 58 | 739 |
| Research Productivity | 240 | 96 | 34 | 294 | 668 |
| Organizational Efficiency | 370 | 82 | 63 | 35 | 553 |
| Technology Adoption Rate | 296 | 118 | 66 | 29 | 513 |
| Firm Productivity | 277 | 34 | 68 | 10 | 394 |
| AI Safety & Ethics | 117 | 177 | 44 | 24 | 364 |
| Output Quality | 244 | 61 | 23 | 26 | 354 |
| Market Structure | 107 | 123 | 85 | 14 | 334 |
| Decision Quality | 168 | 74 | 37 | 19 | 301 |
| Fiscal & Macroeconomic | 75 | 52 | 32 | 21 | 187 |
| Employment Level | 70 | 32 | 74 | 8 | 186 |
| Skill Acquisition | 89 | 32 | 39 | 9 | 169 |
| Firm Revenue | 96 | 34 | 22 | — | 152 |
| Innovation Output | 106 | 12 | 21 | 11 | 151 |
| Consumer Welfare | 70 | 30 | 37 | 7 | 144 |
| Regulatory Compliance | 52 | 61 | 13 | 3 | 129 |
| Inequality Measures | 24 | 68 | 31 | 4 | 127 |
| Task Allocation | 75 | 11 | 29 | 6 | 121 |
| Training Effectiveness | 55 | 12 | 12 | 16 | 96 |
| Error Rate | 42 | 48 | 6 | — | 96 |
| Worker Satisfaction | 45 | 32 | 11 | 6 | 94 |
| Task Completion Time | 78 | 5 | 4 | 2 | 89 |
| Wages & Compensation | 46 | 13 | 19 | 5 | 83 |
| Team Performance | 44 | 9 | 15 | 7 | 76 |
| Hiring & Recruitment | 39 | 4 | 6 | 3 | 52 |
| Automation Exposure | 18 | 17 | 9 | 5 | 50 |
| Job Displacement | 5 | 31 | 12 | — | 48 |
| Social Protection | 21 | 10 | 6 | 2 | 39 |
| Developer Productivity | 29 | 3 | 3 | 1 | 36 |
| Worker Turnover | 10 | 12 | — | 3 | 25 |
| Skill Obsolescence | 3 | 19 | 2 | — | 24 |
| Creative Output | 15 | 5 | 3 | 1 | 24 |
| Labor Share of Income | 10 | 4 | 9 | — | 23 |
Human Ai Collab
Remove filter
Large language models (LLMs) and agentic systems have shown promise for automated software development.
Statement in paper referencing prior successes of LLMs and agentic systems for automated software development (no empirical data reported in this excerpt).
Trained participants more often assigned tasks to the agent by defining strategies compared to participants who did not receive teamwork training.
Behavioral measure in experiment (frequency of assigning tasks using defined strategies) comparing trained vs. untrained participants in the KeyWe game with a scripted agent.
Participants who received the training delegated a higher percentage of tasks to the agent than participants who did not receive teamwork training.
Between-subjects comparison in KeyWe testbed with a scripted agent; measured percentage of tasks delegated by participants in trained vs. untrained groups.
A HAT training intervention that took less than 30 minutes was developed to train humans on seven teamwork competencies.
Study description: developed a training intervention under 30 minutes targeting seven teamwork competencies; implemented as part of the experiment.
The largest gains appear when AI is embedded in an orchestrated workflow rather than deployed as an isolated coding assistant.
Central thesis supported by comparisons across five delivery configurations (traditional baseline and V1–V4) in a retrospective longitudinal field study of the Chiron platform applied to three real software modernization programs; authors observe greater portfolio-level improvements when AI is integrated into coordinated workflows.
V3 and V4 add acceptance-criteria validation, repository-native review, and hybrid human-agent execution, simultaneously improving speed, coverage, and issue load.
Observed differences across the five delivery configurations (baseline, V1–V4) in the field study of three modernization programs; authors link feature additions in V3/V4 to measured improvements in stage durations, coverage, and validation issues.
First-release coverage rises from 77.0% to 90.5% across the portfolio as platform versions progress.
Observed first-release coverage measured in the retrospective longitudinal field study of three real modernization programs, reported as percentages across delivery configurations.
Validation-stage issue load falls from 8.03 to 2.09 issues per 100 tasks across the portfolio as platform versions progress.
Observed outcomes from the retrospective field study on three programs; validation-stage issues counted and normalized per 100 tasks across delivery configurations.
Modeled senior-equivalent effort falls from 1080.0 to 139.5 SEE-days under the platform configurations studied.
Modeled senior-equivalent effort computed from the study's staffing scenarios and observed outputs across the three real programs.
Modeled raw effort falls from 1080.0 to 232.5 person-days under the platform configurations studied (baseline -> V4 aggregate).
Modeled outcomes computed from observed task volumes and explicit staffing scenarios in the retrospective longitudinal field study covering three real programs.
Portfolio totals move from 36.0 to 9.3 summed project-weeks under baseline staffing assumptions (across the three studied programs and five delivery configurations).
Retrospective longitudinal field study of the Chiron platform applied to three real software modernization programs (COBOL banking migration ~30k LOC, accounting modernization ~400k LOC, .NET/Angular mortgage modernization ~30k LOC); observed and modeled outcomes were aggregated to produce portfolio totals under explicit staffing scenarios.
Because instructional signals are usable only when the learner has acquired the prerequisites needed to parse them, the effective communication channel depends on the learner's current state of knowledge and becomes more informative as learning progresses.
Theoretical consequence derived from the model's prerequisite-structure assumption and sequential teaching formalization (as described in the abstract).
Generative AI has transformed the economics of information production, making explanations, proofs, examples, and analyses available at very low cost.
Statement in paper (intro/abstract) asserting an empirical/observational fact about generative AI; no empirical sample or data reported in the abstract.
These results highlight the importance of trustworthy AI mediation tools in contexts where not only truth, but also trust and confidence matter.
Policy/recommendation based on experimental findings that AI mediation lowers perceived trust and confidence even when accuracy is unchanged.
Reinforcement learning (post-training) on our corpus improves downstream embodied manipulation performance.
Downstream evaluation described in the paper showing improved performance on embodied manipulation tasks after RL post-training on MultihopSpatial-Train.
Reinforcement learning (post-training) on our MultihopSpatial-Train corpus enhances intrinsic VLM spatial reasoning.
Experimental intervention: RL-based post-training on the authors' training corpus followed by evaluation on intrinsic spatial reasoning benchmarks (described in the paper).
We provide MultihopSpatial-Train, a dedicated large-scale training corpus intended to foster spatial intelligence in VLMs.
Dataset/resource contribution described in the paper (existence and intended use of MultihopSpatial-Train).
We propose Acc@50IoU, a complementary metric that simultaneously evaluates reasoning and visual grounding by requiring both answer selection and precise bounding box prediction.
Methodological contribution in the paper defining the Acc@50IoU metric and its intended use to measure combined answer correctness and bounding-box IoU >= 0.5.
We introduce MultihopSpatial, a comprehensive benchmark designed for multi-hop and compositional spatial reasoning, featuring 1- to 3-hop complex queries across diverse spatial perspectives.
Dataset/benchmark construction described in the paper (design and scope of MultihopSpatial).
Spatial reasoning is foundational for Vision-Language Models (VLMs), particularly when deployed as Vision-Language-Action (VLA) agents in physical environments.
Conceptual/introductory statement in the paper motivating the work (literature-based argument about VLMs and VLA agents).
An approach is needed focused on emerging and future interdependencies between professionals and generative machine learning, implying extending but also reimagining theoretical perspectives on expertise, work and organizations.
Paper's central argument based on theoretical reasoning and literature synthesis about generative ML characteristics and their implications for professionals; method: conceptual/theoretical development; no empirical sample.
Existing theories need to be extended whilst also responding to the distinctive characteristics of generative machine learning and the implications for how we theorize change.
Argumentative/theoretical claim in the paper based on comparison of features of generative ML with prior digital/algorithmic technologies; method: conceptual analysis and literature engagement; no empirical sample.
We develop an approach using insights from existing literature on digital, algorithmic and artificial intelligence technologies.
Paper's stated contribution: theoretical development based on synthesis of existing literature (digital, algorithmic, AI). Method: conceptual synthesis; no empirical testing or sample reported.
There is a need for an approach to theorizing professional work and professional service firms in the generative machine learning age.
Conceptual argument presented in the paper (literature-based rationale); method is theoretical/literature review and argumentation; no empirical sample reported.
The technology particularly benefits less experienced practitioners by providing comprehensive starting points for legal research, while experienced attorneys can use it for quality control and initial drafts.
Authors' interpretation of AI outputs from the experiment and reasoning about how those outputs map onto different practitioner needs (qualitative judgment).
The analysis reveals AI’s potential to transform law firm economics by dramatically reducing research time while maintaining analytical quality, though careful attorney oversight remains essential.
Inference from the experimental finding that four AI systems produced substantive analysis comparable to junior-associate work on one transcript and the stated observation about traditional research time (8–40 hours); authors' qualitative judgment about economic implications and need for oversight.
Statutory and regulatory citations proved generally accurate and useful.
Authors' examination of statutory and regulatory references produced by the four AI engines in the experiment, judged to be generally correct and helpful.
All four engines successfully spotted legal issues, assessed claim strengths and weaknesses, and suggested follow-up investigation—tasks that traditionally required eight to forty hours of junior attorney research time.
Observed outputs from the four AI engines on the single transcript showing issue-spotting, strengths/weaknesses assessment, and suggested follow-ups; comparison to typical junior attorney research time (stated as 8–40 hours).
Contemporary generative AI performs sophisticated legal analysis comparable to experienced associates, correctly identifying major employment law claims including ADA violations, Title VII discrimination, OSHA retaliation, FMLA interference, and workers’ compensation retaliation.
Qualitative assessment of outputs from the four AI engines applied to the single hypothetical transcript; comparison against expected legal claims (authors' judgment that outputs matched those an experienced associate would produce).
Four major generative AI engines—DeepSeek, Claude, ChatGPT, and Grok—are useful legal analysis tools for employment law practitioners.
Experimental evaluation in which a single hypothetical client interview transcript was submitted to each of the four AI systems and their outputs were assessed by the authors.
Organizational support and continuous learning are important to maximize the benefits of AI integration in startup environments.
Conclusions drawn from thematic analysis of interviews with 12 startup employees emphasizing need for organizational support and ongoing learning.
AI functions as a workforce augmentation tool that enhances human capabilities rather than replacing employees.
Reported perceptions from 12 startup employees in semi-structured interviews; thematic coding indicated view of AI as augmentation rather than replacement.
Most employees demonstrated progressive adjustment and competency improvement over time after initial adaptation.
Interview data from 12 startup employees with thematic analysis indicating progressive adjustment and competency gains over time.
AI improves employee performance by supporting more accurate decision-making and increasing work effectiveness and output quality.
Findings from semi-structured interviews of 12 startup employees, analyzed via thematic coding and frequency scoring, reporting improved decision accuracy and output quality with AI support.
AI integration contributes to competency development, particularly in digital literacy, analytical thinking, and adaptive learning.
Qualitative semi-structured interviews with 12 startup employees; thematic coding highlighted competencies (digital literacy, analytical thinking, adaptive learning).
AI significantly enhances employee productivity by accelerating task completion, reducing manual workload, and improving workflow efficiency.
Qualitative study using semi-structured interviews with 12 startup employees; data analyzed with thematic coding, frequency scoring, and visualized analysis.
Structured intent representations (PPS) can improve alignment and usability in human–AI interaction, especially in tasks where user intent is inherently ambiguous.
Synthesis of experimental findings (rendered PPS better on goal_alignment overall, task-dependent gains concentrated in high-ambiguity business tasks) and the preliminary user survey.
A preliminary retrospective survey (N = 20) suggests a 66.1% reduction in follow-up prompts required, from 3.33 to 1.13 rounds, when using PPS.
Authors report a small retrospective survey of N = 20 respondents comparing number of follow-up prompt rounds required before vs after adopting PPS (self-reported).
We introduce goal_alignment, a user-intent-centered evaluation dimension, and find that natural-language-rendered PPS outperforms both simple prompts and raw PPS JSON on this metric.
Experimental comparison across the three prompt conditions using the goal_alignment evaluation dimension applied to the collected outputs (540 outputs across 60 tasks and 3 models), as judged by an LLM judge.
A mixed-methods empirical research agenda is presented, proposing a future PLS-SEM approach to test the mediating role of the cognitive flywheel and the moderating effect of fractal governance on organizational resilience.
Methodological proposal described in the paper (research design and proposed analytic approach); no executed empirical study or sample reported.
Fractal governance architecture is proposed to mitigate systemic vulnerabilities such as automation bias.
Conceptual proposal of a governance design in the paper; no empirical test or sample provided.
The cognitive flywheel is the central mechanism of this dynamic capability and can be operationalized (the paper operationalizes the cognitive flywheel).
Theoretical operationalization within the paper (concept definition and proposed operational measures); no empirical measurement or sample reported.
The co-evolutionary dynamic is formalized using coupled non-linear differential equations and time decay integrals.
Mathematical formalization reported in the paper (modeling methods described); no empirical parameter estimation or sample provided.
Dynamic cognitive advantage arises from the historical, recursive, structural coupling of human semantic intent and machine syntactic processing (a co-evolutionary dynamic).
Conceptual theory introduced and argued in the paper (mechanism-level proposition); formalization provided but no empirical validation.
Conceptualizing the enterprise as a complex adaptive system operating far from thermodynamic equilibrium provides a more appropriate framing for organizations integrating AI and enables the theory of dynamic cognitive advantage.
Theoretical development and conceptual argumentation within the paper; formal framing rather than empirical test; no sample reported.
We propose a multi-agent discussion framework wherein specialized agents collaboratively process extensive product information, distributing cognitive load to alleviate single-agent attention bottlenecks and capturing critical decision factors through structured dialogue.
Method description: multi-agent discussion architecture described and implemented; claimed to distribute cognitive load and reduce single-agent attention bottlenecks (design + reported behavior).
To enhance simulation stability, we implement a mean-field mechanism designed to model the dynamic interactions between the product environment and customer populations, effectively stabilizing sampling processes within high-dimensional decision spaces.
Method description: implementation of a mean-field mechanism within the simulator; paper asserts this design stabilizes sampling in high-dimensional decision spaces (method + reported simulation behavior).
We introduce a preference learning paradigm in which LLMs are economically aligned via post-training on extensive, heterogeneous transaction records across diverse product categories.
Method description: post-training LLMs on heterogeneous transaction records across product categories to align preferences (methodological / training procedure described).
This paper introduces a Multi-Agent Large Language Model-based Economic Sandbox (MALLES) as a unified simulation framework applicable to cross-domain and cross-category scenarios.
Paper description: design and implementation of MALLES, presented as a unified framework leveraging large-scale LLM generalization for cross-domain/cross-category simulation (methodological contribution).
Leaders' AI symbolization lessens AI's negative impact on employees' emotional exhaustion.
Moderation analysis in the four-stage longitudinal study of 285 finance professionals; leader AI symbolization tested as moderator of AI usage -> emotional exhaustion path.