Towards Sustainable Community-Designed AI Systems in the Public Sector

COMPASS 2025 Workshop: July 22/23, 2025

Advancements in data-driven modeling and artificial intelligence applications in public sector settings can lead to increased tensions between stakeholder needs and tangible model capabilities from developers. Engaging multiple stakeholder groups in model development through the gleaning of sociotechnical, theoretical insights using qualitative methods can contribute to more impactful, beneficial model solutions. Qualitative methods can complement quantitative model development, emphasizing stakeholder priorities, informing model design choices, and promoting sustainable connections and model longevity. We are excited to discuss human-centered qualitative methods, sustainable modeling practices, and community engagement in this workshop, engaging with scholars on system-level modeling strategies.

Our key objectives are to:

  • Co-develop a practical framework for integrating qualitative insights into quantitative model development, drawing on participants’ experiences in health, education, housing, and other public sector domains. This directly addresses our central theme: informing quantitative insights through qualitative methods to meet UN sustainability goals.
  • Surface shared challenges and tensions faced by researchers and practitioners working in applied AI, including stakeholder-developer misalignment, model deployment barriers, and limitations of existing governance tools.
  • Define sustainable practices for public sector AI development, including environmental sustainability, long-term stakeholder engagement, and governance responsiveness.
  • Foster interdisciplinary connections across academic and applied contexts, connecting researchers working in SIGCHI spaces with community members and public sector partners, with the goal of building an ongoing network.
  • Generate living resources for participants’ future projects. We also plan to initiate a communication channel to support continued collaboration and iteration post-workshop.

Join us at COMPASS 2025 in Toronto, Canada! We will examine how community engagement and qualitative exploratory methods can inform qualitative model development for public sector AI. As researchers in human-centered AI, the involvement of stakeholders in health, housing, child welfare, social assistance, and education is important to resolve domain and developer conflicts, integrate sustainable practices, and engage community members in model development. Our themes include:

  • Community Engagement: Developing system-level public interest AI for stakeholders, with stakeholders
  • System-Level Insights: Integrating qualitative insights into quantitative model development of local AI
  • Sustainability: Defining sustainable practices when informing quantitative, human-centered practices

We invite researchers, practitioners, providers, professionals, and students working in AI, as well as community members, to join us as we explore qualitative and quantitative insights when engaging community throughout AI development. We ask interested participants to submit a 250-word paragraph to hcds.uoft@gmail.com (with the subject “COMPASS workshop”), describing:

  1. Your research interests (field, research site)
  2. Your research methods (qualitative and/or quantitative)
  3. Your motivation for attending this workshop

Registration information for COMPASS can be found here. We look forward to collaborating with you soon!