Towards Openness Beyond Open Access: User Journeys through 3 Open AI Collaboratives

Open AI collaboratives democratize AI development beyond major tech companies: The three communities studied (BigScience, The Turing Way, and Mozilla's Building Trustworthy AI group) demonstrate that meaningful AI work—from model building to tool development to ecosystem creation—can be accomplished through distributed, volunteer-led teams rather than requiring massive corporate resources.

Different community structures serve different purposes: Each collaborative uses distinct organizational approaches tailored to their goals, suggesting there is no one-size-fits-all model for open AI work. The paper documents how these varying collaboration styles enable communities to effectively pursue their specific objectives.

"Openness" extends beyond just sharing code: The research reveals that open AI collaboratives practice a "broader practice of openness" that goes beyond traditional open-source access, encompassing inclusive participation pathways, transparent decision-making processes, and diverse ways of contributing to AI development.

Clear user journeys help sustain volunteer participation: By mapping how members discover, join, contribute, and participate in these communities, the paper highlights that successful open AI collaboratives intentionally design pathways for engagement—suggesting that accessibility and clear entry points are critical for building diverse, sustainable teams.

Open collaboratives bring workforce diversity to AI: The paper emphasizes "the diversity of AI work and workers" that emerges through these communities, implying that open models may help address representation gaps in AI development by enabling participation beyond traditional tech industry hiring.