Funding AI for Good: A Call for Meaningful Engagement

Paper Title

Funding AI for Good: A Call for Meaningful Engagement

Publication Info

• Topic area: Analysis of funding agendas in Artificial Intelligence for Social Good (AI4SG) initiatives.
• Keywords: AI4SG, funding agendas, techno-centric, balanced approaches, community engagement, HCI, social impact, funding rhetoric, document analysis, interdisciplinary collaboration.

Background and Problem

• Problem / challenge: AI4SG initiatives often fail to deliver tangible, sustainable benefits due to biases in funding agendas that prioritize technological capacities over contextual understanding and community engagement.
• Significance: Funding agendas play a critical role in shaping AI4SG projects, influencing their design, implementation, and outcomes. Addressing biases in funding rhetoric can improve the alignment of AI4SG initiatives with their intended social impact goals.
• Motivation and related work: Prior work has focused on the challenges faced by AI4SG project teams and the outcomes of funded projects but has largely overlooked the upstream influence of funding agendas. Existing literature highlights the importance of balancing technical and contextual considerations for impactful innovations, but funding documents often promote a techno-centric approach.

Solution

• Proposed approach: A reflexive thematic analysis of 35 AI4SG funding documents to uncover how funding rhetoric frames AI4SG and influences downstream project approaches.
• Novelty:
  1. First qualitative analysis of AI4SG funding documents as discourse instruments rather than neutral texts.
  2. Identification of a spectrum from techno-centric to balanced approaches in funding rhetoric.
  3. Recommendations for designing funding calls to better align with community needs and social impact goals.
• Procedure and key techniques:
  • Collection of 35 funding documents representing $410 million in investments from 23 funders.
  • Reflexive thematic analysis to identify themes in four components: background, expected outcomes, eligibility criteria, and funders’ support.
  • Characterization of themes along a spectrum from techno-centric to balanced approaches.

Results

• Concrete findings:
  • Many funding documents emphasized AI’s transformative potential without sufficient contextual grounding.
  • Expected outcomes often prioritized technical deliverables over community benefits.
  • Eligibility criteria frequently favored applicants with existing AI capacities, limiting community involvement to consultative roles.
  • Post-deployment funding support and training for community engagement were notably lacking.
• Advantage over baselines:
  • Identification of funding documents that adopt more balanced approaches, such as requiring domain expertise, emphasizing community outcomes, and partnering with regional organizations.
• Experiments / evaluation:
  • Analysis of funding documents from diverse funders, including government agencies, private tech companies, and philanthropic foundations.
  • Use of qualitative coding and affinity diagramming to uncover thematic patterns.
• Limitations and future work:
  • Dataset limited to English-language documents and publicly available materials.
  • Need for further research linking funding rhetoric to project outcomes and expanding analysis to non-Western funders.

Summary

This study examines how funding documents shape AI4SG initiatives, revealing a spectrum from techno-centric to balanced approaches. It identifies dissonances between funders’ intentions for social impact and the techno-centric rhetoric in some funding calls. Recommendations are provided for designing funding calls that emphasize community engagement, contextual understanding, and sustainable outcomes. The findings highlight opportunities for HCI researchers and practitioners to mediate between funders and communities, fostering more equitable and impactful AI4SG initiatives.