Summary
This survey came about to inform the shape of the Movement for a Better Internet policy lab on the overarching topic of Generative Artificial Intelligence & the Future of Creativity. The topic with the highest level of interest for respondents is Generative AI, Creativity & the Commons; this topic is also the highest priority for respondents. As a result, this topic will be a focus of the upcoming policy lab.
Full Report
Purpose
This survey came about to inform the shape of the Movement for a Better Internet’s first policy lab, on the overarching topic of Generative Artificial Intelligence & the Future of Creativity. The survey was sent to all members of the movement in order to understand their areas of focus and priorities for policy interventions.
Specifically, this survey was designed to learn from the movement membership, both their organization’s level of interest in, and prioritization of these five subtopics:
Consent & Permission to Train AI Models on Artistic Works
Competition & Impacts on Working Artists
Credit and Attribution for AI-Created Works
Culture and/or/versus Commerce
Generative AI, Creativity & the Commons
Response and outcomes
The survey garnered a ~15% response rate from among the organizations that have elected to receive information from the movement. We also received some insightful open-ended comments to questions asking about other important topics on the subject of generative AI and the future of creativity and for anything else the organizing partners should consider while moving forward.
The topic with the highest level of interest for respondents is Generative AI, Creativity & the Commons; this topic is also the highest priority for respondents. As a result, this topic will be the primary focus of the upcoming policy lab:
Generative AI & the Commons: How might policy ensure that generative AI contributes to a thriving commons of widely accessible knowledge and creativity that people may build upon? How do we address concerns about the impoverishment of the commons due to creators no longer sharing their works publicly on the Web to avoid AI training? What impacts can we foresee on openly licensed content and public interest initiatives; for example if people can use ChatGPT to get answers gleaned from Wikipedia without ever visiting Wikipedia, will Wikipedia’s commons of information continue to be sustainable? Exploration here might consider options noted above, as well as other forms of remuneration and public funding schemes.
Subtopic suggestions
Below are the paraphrased responses to the open ended question:
Is there another topic, not listed above, under the overarching topic of Generative AI & the Future of Creativity that you would like to share for consideration?
Supporting educators in leveraging GAI to develop high quality Open Educational Resources and how to do this responsibly.
The need for transparency on the coding of models and the data sets on which they are trained.
Enhancing public trust by way of enabling transparency and an “open” model for AI which can lead to fostering innovation and economic activity, stimulating creative and cultural use, and democratizing access.
Open infrastructure (as opposed to closed and for-profit infrastructure) as a public utility and means of generating public goods, and creative tools for addressing social issues.
Focus on equitable access to, and use of tech in nonprofits and changemaking initiatives.
How organizations can contribute to inequitable impacts on artists by their use of AI tools and how to support artists in the community to prevent it.
Privacy and human rights data needs to be considered as a means of ensuring that artists seeking social justice are not subject to surveillance and censorship from repressive regimes. Informed consent models are another way of ensuring that artists and their work are not exploited.
Anything else the organizers should know for the policy lab?
Below are the paraphrased responses to the open ended question:
Is there anything else you or your organization would like other members of the Movement for a Better Internet to know concerning the upcoming policy lab in regard to Generative AI & the Future of Creativity?
The current focus on the use of copyrighted works to train models may be distracting us from the real risk: commercialization of outputs by corporate rights holders. We need to move communities to understand that the impact of technology [concerning AI] is going to be in the outputs, and they need to be thinking about how that impacts them and how to get ahead of it in the policy space.
It is important to address generativeAI in educational spaces and there is already a great deal of work being done to realize its potential.
We need a guide for organizing local events about the Better Internet movement.
We need to be more proactive in influencing how AI is playing out, and it will take a whole ecosystem approach.
Include global perspectives.
Appendix
This appendix contains the actual text of the questions asked in the survey, charts summarizing the responses, and the actual and paraphrased text responses to the two open-ended questions.
Respondent organizations’ level of interest in each sub-topic
The question
We would like to learn about your organization’s level of interest in discussing each of the subtopics listed under the overarching topic of generative AI. This will help us understand which of the subtopics are of most interest to member organizations and how to focus the policy lab.
Please rate your organization's interest in each topic according to this scale:
1 = no interest in exploring the policy implications of this topic
2 = a little interested in exploring the policy implications of this topic
3 = neutral on the topic / don't have an opinion
4 = interested in exploring the policy implications of this topic
5 = very interested/feel it is essential to discuss the policy implications of this topic
Summarized responses:
Consent & Permissions to Train AI Models on Artistic Works
How might policy reconcile the interests of content creators, AI firms, and users in the training of generative AI systems? What mechanisms might help people signal preferences and secure permission for AI training? How might such mechanisms encompass the diversity of types of generative AI and many different types of uses? Exploration here might include technical measures for signaling preferences, development of databases of works permissioned for AI training and addressing how AI is used to mimic artists’ work.
Figure 1. Consent & Permissions to Train AI Models on Artistic Works
Member organizations’ level of interest in discussing “Consent & Permissions to Train AI Models on Artistic Works”.
Competition & Impacts on Works
How might policy address the risk of unfair competition or professional dislocation for existing artists? How best to ensure that tools’ benefits are well distributed, supporting working artists, and not simply have the benefits accrue to those creating AI tools? Are there ways to ensure generative AI is used more to augment existing labor rather than automate and replace it? Exploration here might include a focus on labor organizing, tax and social safety net policy, and social funding for the arts.
Figure 2. Competition & Impacts on Works.
Member organizations’ level of interest in discussing “Competition & Impacts on Works”.
Credit and Attribution for AI-Created Works
How might policy help evolve norms and practices around attribution in this new space? What technical or other mechanisms might help, recognizing that all creativity builds on the past? Exploration here might focus on how norms of attribution have built in a variety of areas, as well as technical solutions for mapping attribution.
Figure 3. Credit & Attribution for AI-Created Works.
Member organizations’ level of interest in discussing “Credit & Attribution for AI-Created Works”.
Culture and/or/versus Commerce
As generative AI enters the picture, how might policy return to or reimagine the ongoing debate about commercialism and media consolidation’s effects on the arts? What interventions are needed to support a thriving, diverse artistic culture?
Figure 4. Culture and/or/versus Commerce.
Member organizations’ level of interest in discussing “Culture and/or/versus Commerce”.
Generative AI, Creativity & the Commons
How might policy ensure that generative AI contributes to a thriving commons of widely accessible knowledge and creativity that people may build upon? How do we address concerns about the impoverishment of the commons due to creators no longer sharing their works publicly on the Web to avoid AI training? What impacts can we foresee on openly licensed content and public interest initiatives; for example if people can use ChatGPT to get answers gleaned from Wikipedia without ever visiting Wikipedia, will Wikipedia’s commons of information continue to be sustainable? Exploration here might consider options noted above, as well as other forms of remuneration and public funding schemes.
Figure 5. Generative AI, Creativity & the Commons.
Member organizations’ level of interest in discussing “Generative AI, Creativity & the Commons”.
How respondents prioritize each of the the subtopics
The question
Please rank how your organization prioritizes the subtopics listed below relative to each other. Refer to the descriptions of the other topics above as needed.
Figure 6. Screenshot of Form Illustrating Ranking Matrix.
A screenshot of the form sent to members asking them to prioritize the subtopics relative to each other.
Summary of responses
Figure 7. Summary of responses prioritizing subtopics relative to each other.
