FROM PRINCIPLES TO PRACTICE: PARTICIPATORY INFRASTRUCTURES FOR INCLUSIVE AI

Abstract

As AI technologies become more integrated and embedded into everyday services, activities, and social connectivity, the notion of inclusive AI is increasingly pertinent; yet the concept lacks theoretical cohesion and remains elusive in practice. In this paper we propose a practice-based framework for inclusive AI, developed through collaborations with community, humanitarian, health, legal, and learning organisations. We conceptualise inclusive AI not as a technological output but as a participatory process, situating inclusion at the centre of capability building, equitable benefit, and collective participation. Central to this reframing is the role of participatory infrastructures and intermediaries—individuals, organisations, and public or online spaces—who act as capability converters, translating technical features into community-relevant outcomes. Our evaluative framework extends responsible AI by looking beyond principles and technical fixes, to also asking whether systems enable communities to achieve tangible benefits and equitable participation across design, deployment, and governance. It identifies four dimensions of inclusive AI: safety and accountability, community participation, community benefit, and sustainability and maintenance. We argue that achieving inclusivity requires infrastructures of translation, adaptation, and care, supported by intermediaries who are both technically informed and culturally situated. The framework offers researchers and practitioners a conceptual and evaluative tool to embed inclusivity throughout the AI lifecycle, ensuring systems are not only technically sound but socially meaningful.