Inclusive.AI: Towards Democratic AI with DAO-Enabled Inclusive Decision-Making

A major criticism of AI development is the lack of transparency, such as, inadequate documentation and traceability in its design and decision-making processes, leading to adverse outcomes including discrimination, lack of inclusivity and representation, and breaches of legal regulations. Underserved populations, in particular, are disproportionately affected by these design decisions. Furthermore, traditional social science techniques such as interviews, focus groups, and surveys struggle to adequately capture user needs and expectations in the digital era, due to their inherent limitations in deliberation, consensus-building, and providing consistent insights. We developed Inclusive.AI, a democratic platform utilizing Decentralized Autonomous Organization (DAO) mechanisms to enable underserved groups to deliberate and reach a consensus on key AI issues, such as how to address stereotypical biases in text-to-image models. We designed and evaluated various DAO configurations, such as voting schemes, to support democratic decision-making in AI governance. Through a 2x2 experimental design, we manipulated voting methods (ranked vs. quadratic) and power distribution (equal vs. 20/80 differential) in a randomized online experiment (n=177) with participants from the Global South and people with disabilities, to study the impact of these conditions on people’s perceptions on the AI model decision-making processes. Our results suggest that despite their diverse backgrounds, participants showed convergence in deliberations on several aspects, including user control over image generation, multiple output options for user selection, and the social appropriateness and accuracy of generated images. Additionally, different voting configurations led to varied outcomes in AI model update decisions. Notably, the combination of quadratic voting—which gives minority more voice—and equal power distribution is perceived as a more fair and democratic approach.

Files

Metadata

Work Title Inclusive.AI: Towards Democratic AI with DAO-Enabled Inclusive Decision-Making
Access
Open Access
Creators
  1. Tanusree Sharma
  2. Yujin Potter
  3. Jongwon Park
  4. Yiren Liu
  5. Yun Huang
  6. Sunny Liu
  7. Dawn Song
  8. Jeff Hancock
  9. Yang Wang
Keyword
  1. AI Governance
  2. Decentralization
  3. Decentralized Autonomous Organizations
  4. Responsible AI
License In Copyright (Rights Reserved)
Work Type Research Paper
Acknowledgments
  1. OpenAI
Publication Date 2024
Deposited December 22, 2024

Versions

Analytics

Collections

This resource is currently not in any collection.

Work History

Version 1
published

  • Created
  • Updated
  • Updated Keyword, Description, Publication Date Show Changes
    Keyword
    • AI Governance, Decentralization, Decentralized Autonomous Organizations, Responsible AI
    Description
    • A major criticism of AI development is the lack of transparency, such as, inadequate documentation and traceability in its design and decision-making processes, leading to adverse outcomes including discrimination, lack of inclusivity and representation, and breaches of legal regulations. Underserved populations, in particular, are disproportionately affected by these design decisions. Furthermore, traditional social science techniques such as interviews, focus groups, and surveys struggle to adequately capture user needs and expectations in the digital era, due to their inherent limitations in deliberation, consensus-building, and providing
    • consistent insights. We developed Inclusive.AI, a democratic platform utilizing Decentralized Autonomous Organization (DAO) mechanisms to enable underserved groups to deliberate and reach a consensus on key AI issues, such as how to address stereotypical biases in text-to-image models. We designed and evaluated various DAO configurations, such as voting schemes, to support democratic decision-making in AI governance. Through a 2x2 experimental design, we manipulated voting methods
    • (ranked vs. quadratic) and power distribution (equal vs. 20/80 differential) in a randomized online experiment (n=177) with participants from the Global South and people with disabilities, to study the impact of these conditions on people’s perceptions on the AI model decision-making processes. Our results suggest that despite their diverse backgrounds, participants showed convergence in deliberations on several aspects, including user control over image generation, multiple output options for user selection, and the social appropriateness and accuracy of generated images. Additionally, different voting configurations led to varied outcomes in AI model update decisions. Notably, the combination of quadratic voting—which gives minority more voice—and equal power distribution is perceived as a more fair and democratic approach.
    Publication Date
    • 2024
  • Updated Acknowledgments Show Changes
    Acknowledgments
    • OpenAI
  • Added Creator Tanusree Sharma
  • Added InclusiveAI_Preprint.pdf
  • Updated Subtitle, License Show Changes
    Subtitle
    • Engaging Underserved Populations in Democratic Decision-Making on AI
    License
    • https://rightsstatements.org/page/InC/1.0/
  • Published
  • Updated

Version 2
published

  • Created
  • Added Creator Yujin Potter
  • Added Creator Jongwon Park
  • Added Creator Yiren Liu
  • Added Creator Yun Huang
  • Added Creator Sunny
  • Added Creator Dawn Song
  • Added Creator Jeff Hancock
  • Added Creator Yang Wang
  • Published
  • Updated
  • Updated Subtitle Show Changes
    Subtitle
    • Engaging Underserved Populations in Democratic Decision-Making on AI
  • Renamed Creator Sunny Liu Show Changes
    • Sunny
    • Sunny Liu