Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Foregrounding Artist Opinions: A Survey Study on Transparency, Ownership, and Fairness in AI Generative Art

Juniper Lovato, Julia Zimmerman, Isabelle Smith, Peter Dodds, Jennifer Karson

TL;DR

The paper investigates how artists perceive Generative AI art, focusing on utility, threats to livelihoods, data disclosure, derivative ownership, and fair compensation. It uses a two-phase survey (N=459) and applies ordinal logistic regression to identify associations between opinions and respondent characteristics, revealing strong support for training-data disclosure, divided views on ownership, and nuanced concerns about profits and workforce impact. The findings underscore a demand for transparency and equitable practices, while showing a spectrum of attitudes toward AI as a positive development. This work contributes actionable guidance for AI researchers and industry to align GenAI practices with artists' values and to inform policy and design choices that protect creative labor.

Abstract

Generative AI tools are used to create art-like outputs and sometimes aid in the creative process. These tools have potential benefits for artists, but they also have the potential to harm the art workforce and infringe upon artistic and intellectual property rights. Without explicit consent from artists, Generative AI creators scrape artists' digital work to train Generative AI models and produce art-like outputs at scale. These outputs are now being used to compete with human artists in the marketplace as well as being used by some artists in their generative processes to create art. We surveyed 459 artists to investigate the tension between artists' opinions on Generative AI art's potential utility and harm. This study surveys artists' opinions on the utility and threat of Generative AI art models, fair practices in the disclosure of artistic works in AI art training models, ownership and rights of AI art derivatives, and fair compensation. Results show that a majority of artists believe creators should disclose what art is being used in AI training, that AI outputs should not belong to model creators, and express concerns about AI's impact on the art workforce and who profits from their art. We hope the results of this work will further meaningful collaboration and alignment between the art community and Generative AI researchers and developers.

Foregrounding Artist Opinions: A Survey Study on Transparency, Ownership, and Fairness in AI Generative Art

TL;DR

The paper investigates how artists perceive Generative AI art, focusing on utility, threats to livelihoods, data disclosure, derivative ownership, and fair compensation. It uses a two-phase survey (N=459) and applies ordinal logistic regression to identify associations between opinions and respondent characteristics, revealing strong support for training-data disclosure, divided views on ownership, and nuanced concerns about profits and workforce impact. The findings underscore a demand for transparency and equitable practices, while showing a spectrum of attitudes toward AI as a positive development. This work contributes actionable guidance for AI researchers and industry to align GenAI practices with artists' values and to inform policy and design choices that protect creative labor.

Abstract

Generative AI tools are used to create art-like outputs and sometimes aid in the creative process. These tools have potential benefits for artists, but they also have the potential to harm the art workforce and infringe upon artistic and intellectual property rights. Without explicit consent from artists, Generative AI creators scrape artists' digital work to train Generative AI models and produce art-like outputs at scale. These outputs are now being used to compete with human artists in the marketplace as well as being used by some artists in their generative processes to create art. We surveyed 459 artists to investigate the tension between artists' opinions on Generative AI art's potential utility and harm. This study surveys artists' opinions on the utility and threat of Generative AI art models, fair practices in the disclosure of artistic works in AI art training models, ownership and rights of AI art derivatives, and fair compensation. Results show that a majority of artists believe creators should disclose what art is being used in AI training, that AI outputs should not belong to model creators, and express concerns about AI's impact on the art workforce and who profits from their art. We hope the results of this work will further meaningful collaboration and alignment between the art community and Generative AI researchers and developers.
Paper Structure (20 sections, 2 equations, 7 figures, 36 tables)

This paper contains 20 sections, 2 equations, 7 figures, 36 tables.

Table of Contents

  1. Introduction
  2. Background
  3. Generative AI Models
  4. Related Work
  5. Economic Context: Livelihood of the Artist
  6. Problems in the Pipeline of Generative AI Tools
  7. Methods
  8. Survey Methodology
  9. Data
  10. Analytical Methods
  11. Results
  12. Discussion
  13. Acknowledgements
  14. Appendix A: Survey Questions
  15. Appendix B: Full Table of Descriptive Statistics
  16. ...and 5 more sections

Figures (7)

  • Figure 1: "Art is the joint responsibility of all [people]" larousse1970ancient.
  • Figure 2: The ordinal logistic regression results show the association between answering 'Agree' or 'Strongly Agree' to RQ1 (Generative AI art models as a threat to art workers) and other survey variables. Results are represented by odds ratios on a log scale on the x-axis, with bars conveying a 95% confidence interval and red indicating a p-value $<$ 0.05 and a confidence interval that does not cross the threshold of the dotted line in the figure.
  • Figure 3: The ordinal logistic regression results show the association between answering 'Agree' or 'Strongly Agree' to RQ2 (Generative AI art models as a positive development) and survey secondary variables. Results are represented by odds ratios on a log scale on the x-axis, with bars conveying a 95% confidence interval and red indicating a p-value $<$ 0.05 and a confidence interval that does not cross the threshold of the dotted line in the figure.
  • Figure 4: The ordinal logistic regression results show the association between answering 'Agree' or 'Strongly Agree' to RQ3 (Should model creators be required to disclose in detail what art & images they use to train their AI models) and survey secondary variables. Results are represented by odds ratios on a log scale on the x-axis, with bars conveying a 95% confidence interval and red indicating a p-value $<$ 0.05 and a confidence interval that does not cross the threshold of the dotted line in the figure.
  • Figure 5: The ordinal logistic regression results show the association between answering 'Agree' or 'Strongly Agree' to RQ4A (If an AI art model was used by someone else to produce artwork recognizably in your style (e.g., in the style of Van Gogh) Should that work and its derivatives be considered the property of the person who used the AI model to generate the artwork?) and survey secondary variables. Results are represented by odds ratios on a log scale on the x-axis, with bars conveying a 95% confidence interval and red indicating a p-value $<$ 0.05 and a confidence interval that does not cross the threshold of the dotted line in the figure.
  • ...and 2 more figures