Participatory AI Justice in HCI: A Scoping Review

Paper Title

Participatory AI Justice in HCI: A Scoping Review

Publication Info

• Topic area: Participatory design in AI with a focus on justice and social equity.
• Keywords: Participatory AI, social justice, HCI, design justice, power dynamics, community engagement, AI ethics, co-design, marginalization, AI lifecycle.

Background and Problem

• Problem / challenge: Participatory AI (PAI) practices often fail to address systemic power imbalances and justice issues, leading to tokenistic or performative engagements. Existing knowledge on justice-oriented PAI is fragmented, making it difficult for researchers and practitioners to navigate and implement effectively.
• Significance: Addressing these gaps is crucial to ensure AI systems are inclusive, equitable, and aligned with human values, particularly for marginalized communities disproportionately affected by AI's adverse impacts.
• Motivation and related work: Participatory design has historically aimed to democratize technology development, but its application in AI is fraught with challenges, including resource constraints, corporate co-optation, and epistemic divides. While there is growing interest in justice-oriented PAI, the field lacks a comprehensive, actionable framework for integrating social justice principles into AI design.

Solution

• Proposed approach: A scoping review of 26 articles from HCI venues to map the intersection of participatory AI and social justice, guided by the Design Justice Principles.
• Novelty:
  1. Provides a holistic understanding of justice-oriented PAI practices in HCI.
  2. Offers methodological reflections on the roles of researchers and communities in PAI.
  3. Highlights the potential of artefacts as "AI boundary objects" to mediate participation and contestation.
  4. Advocates for long-term alliances with community-led initiatives to sustain justice-oriented PAI efforts.
• Procedure and key techniques:
  • Conducted a scoping review using the ACM Digital Library, focusing on articles published between 2019 and 2025.
  • Applied deductive thematic analysis based on the Design Justice Principles.
  • Categorized findings into six themes: purpose (why), object of design (what), role of designers (who involves), participants (who is involved), process (how), and lifecycle phase (when).

Results

• Concrete findings:
  • Most PAI work focuses on developing responsible AI, empowering communities, or designing for justice.
  • Participation often occurs in the problem-definition phase of the AI lifecycle.
  • Artefacts like toolkits, mockups, and speculative designs are frequently used to facilitate participation.
  • Researchers' positionality and power dynamics are rarely addressed explicitly.
• Advantage over baselines:
  • Provides a structured, justice-oriented lens for evaluating PAI practices, emphasizing community empowerment and long-term impact.
  • Highlights the dual role of artefacts as tools for engagement and as instruments of contestation.
• Experiments / evaluation:
  • Reviewed 26 articles from leading HCI venues, including CHI, DIS, FAccT, and TOCHI.
  • Analyzed articles based on publication year, geographic affiliation, disciplinary background, and thematic focus.
• Limitations and future work:
  • Limited to published academic articles, potentially excluding valuable non-academic and community-led work.
  • Focused on the Design Justice Principles, which may not capture all dimensions of justice-oriented PAI.
  • Calls for further research on sustaining PAI initiatives and integrating diverse frameworks.

Summary

This paper conducts a scoping review of 26 articles to explore how participatory AI intersects with social justice in HCI. Guided by the Design Justice Principles, it identifies key practices, challenges, and opportunities for justice-oriented PAI. The findings emphasize the importance of early-stage community involvement, the role of artefacts as "AI boundary objects," and the need for long-term alliances with community-led initiatives. While the field shows promise, it remains fragmented, with significant gaps in addressing power dynamics and sustaining engagement. This work provides a foundation for advancing justice-oriented PAI practices and fostering meaningful community empowerment in AI design.