0 cumulative citations
View corpus contextCreative datasets defy blanket data policies: artist-generated materials are heterogeneous and legally fraught, so effective reuse and market-making require bespoke metadata, rights-management, and artist-led governance.
Citation observations
Cumulative provider counts captured on specific dates; providers are never combined.
Creative Data Lab (CDL), hosted as an Advanced Study Group at the Pufendorf Institute for Advanced Studies, Lund University, brought together artist-researchers and research support staff to explore artistic research data management (ARDM) in relation to documentation, storage, sharing, and use. Working across disciplines and in dialogue with national and international debates, the group examined how artistic research data resist fixed definitions, raise questions of ethics and rights, and require infrastructures that can accommodate high-quality, context-dependent materials. The contributions in this publication build on the work initiated in CDL to advance ongoing discussions and support the development of practices and infrastructures on the artistic field’s own terms.
Summary
Main Finding
Artistic research data (ARD) resist one-size-fits-all definitions and standard research-data approaches: they are heterogeneous, context-dependent, ethically and legally fraught, and therefore require bespoke, flexible infrastructures and governance arrangements developed in dialogue with artist-researchers and the artistic field itself.
Key Points
- ARD resists fixed definitions: materials vary in form, purpose, and meaning across artistic practices.
- Ethics and rights are central: questions of authorship, consent, sensitivity, and reuse permeate ARD.
- Context matters: documentation and metadata must capture contextual information to preserve artistic intent and research value.
- Infrastructure gaps: existing research-data systems often cannot accommodate high-quality, non-standard, or multimedia artistic materials.
- Field-led solutions: sustainable practice and infrastructure development should be driven by the needs and norms of the artistic community.
- CDL’s contribution: convening artist-researchers and research support staff produced comparative insights and resources to advance ARDM debates and practice.
Data & Methods
- Organizing format: an Advanced Study Group (Creative Data Lab) hosted at Lund University, combining artist-researchers with research support staff.
- Methods used (as described/implicit): interdisciplinary dialogue, practice-based inquiry, workshops and group deliberation, engagement with national and international policy debates, and compilation of contributed texts that build on the group’s work.
- Output: a publication collecting contributions that synthesize lessons, examples, and recommendations for ARDM practice and infrastructure.
Implications for AI Economics
- Valuation and market design
- Artistic datasets highlight heterogeneity and non-fungibility of data assets; AI-economics models should account for differentiated data quality, provenance, and contextual value rather than treating data as homogeneous goods.
- Transaction costs for cleaning, documenting, and negotiating rights over ARD are high—these inform pricing, contract design, and platform fee structures for specialized dataset markets.
- Incentives and compensation
- Artists’ concerns about rights and reuse point to the need for tailored incentive mechanisms (revenue sharing, attribution, negotiated licenses) when creative labor generates training data for AI.
- Contracting, licensing, and property rights
- ARD demonstrates legal and ethical complexity; AI-economics analyses of data markets must model incomplete contracts, heterogeneous IP regimes, and the role of intermediaries that manage rights and provenance.
- Infrastructure and standards
- Flexible metadata and provenance systems that capture context are economically important: they reduce frictions, increase reuse value, and enable better price discovery for specialized datasets.
- Investments in specialized curation infrastructure (storage, annotation, access controls) are a meaningful cost component and affect returns to data owners and buyers.
- Policy and regulation
- Artist-led governance models and participatory design of data infrastructures are relevant to policy prescriptions: regulation should consider power asymmetries, consent mechanisms, and cultural value preservation.
- Research design for AI economics
- Empirical work should incorporate case studies of non-standard, multimodal datasets (like ARD) to better estimate welfare impacts, market concentration risks, and social surplus from AI systems trained on creative material.
- Modeling should allow for heterogeneous agents (creators, curators, platforms, consumers) and endogenous standards/infrastructure investments.
Practical takeaway: AI economists and policy-makers should treat creative and other context-rich datasets as distinct data classes—requiring bespoke governance, metadata standards, compensation mechanisms, and infrastructure investment—rather than subsuming them under generic data-economy assumptions.
Assessment
Claims (11)
| Claim | Direction | Outcome | Confidence & Evidence | Details |
|---|---|---|---|---|
| Artistic research data (ARD) resist one-size-fits-all definitions because they are heterogeneous, context-dependent, and vary across artistic practices. Organizational Efficiency | mixed | Fit of standardized research-data definitions and approaches to artistic research data |
Reading fidelity
high
Study strength
medium
|
not reported
|
| ARD involve central ethical and legal issues concerning authorship, consent, sensitivity, and reuse. Ai Safety And Ethics | negative | Ethical and legal feasibility of collecting, sharing, and reusing artistic research data |
Reading fidelity
high
Study strength
medium
|
not reported
|
| Documentation and metadata for ARD need to capture contextual information in order to preserve artistic intent and research value. Organizational Efficiency | positive | Preservation of artistic intent and research value through documentation and metadata |
Reading fidelity
high
Study strength
medium
|
not reported
|
| Existing research-data systems often cannot accommodate high-quality, non-standard, or multimedia artistic materials. Organizational Efficiency | negative | Capacity of existing research-data infrastructure to store, manage, and support artistic research materials |
Reading fidelity
high
Study strength
medium
|
not reported
|
| Sustainable ARD practices and infrastructure should be driven by the needs and norms of the artistic community. Governance And Regulation | positive | Sustainability and appropriateness of artistic research-data practices and infrastructure |
Reading fidelity
high
Study strength
medium
|
not reported
|
| The Creative Data Lab's Advanced Study Group generated comparative insights and resources intended to advance debates and practice concerning artistic research data management. Training Effectiveness | positive | Development of knowledge and resources for artistic research-data management |
Reading fidelity
high
Study strength
medium
|
not reported
|
| Creative datasets such as ARD should be treated as heterogeneous and non-fungible data assets rather than as homogeneous goods in AI-economics models. Market Structure | positive | Accuracy of economic models of creative-data valuation and market design |
Reading fidelity
medium
Study strength
speculative
|
not reported
|
| Cleaning, documenting, and negotiating rights over ARD create high transaction costs that should be considered in pricing, contract design, and platform fee structures for specialized dataset markets. Market Structure | negative | Transaction costs and market frictions in specialized artistic-data markets |
Reading fidelity
medium
Study strength
speculative
|
not reported
|
| Rights and reuse concerns in artistic data imply a need for tailored compensation and incentive mechanisms, including revenue sharing, attribution, and negotiated licenses. Wages | positive | Incentives and compensation for artists whose creative labor contributes to AI training data |
Reading fidelity
medium
Study strength
speculative
|
not reported
|
| Flexible metadata and provenance systems that capture context may reduce frictions, increase reuse value, and improve price discovery for specialized datasets. Market Structure | positive | Reuse value, transaction frictions, and price discovery for specialized datasets |
Reading fidelity
medium
Study strength
speculative
|
not reported
|
| AI-economics research should use case studies of non-standard, multimodal datasets such as ARD and model heterogeneous agents and endogenous investments in standards and infrastructure. Research Productivity | positive | Validity of empirical and theoretical analyses of welfare, market concentration, and social surplus from AI systems using creative data |
Reading fidelity
high
Study strength
speculative
|
not reported
|