Bridging AI and Society - Lessons from Participatory Machine Learning Projects

What is it about?

This paper explores how "participation brokers" involve everyday people in developing machine learning (ML) systems. We interviewed 18 brokers to understand their strategies and challenges in facilitating collaboration between ML developers and participants from the broader community. The brokers aim to give participants a voice in shaping ML systems that impact them. However, they often struggle to fit participants' messy, contextual input into the structured data formats needed for ML. Power imbalances between developers and participants also pose challenges. We suggest brokers evolve into community advocates and educators. By empowering participants to envision alternative futures through ML, while also relaying their frustrations, brokers can help steer participatory ML projects to benefit both developers and the public.

Featured Image

Why is it important?

The paper offers unique insights from real-world projects on democratising ML development. It's timely as ML increasingly impacts people's daily lives and broader society, highlighting the need for more inclusive practices.