Participation-washing could be the next dangerous fad in machine learning

Abstraction: Critique of performative participation as insufficient fix for ML injustice

Key points:

• "Participation-washing" mirrors "ethics-washing": superficial inclusion that legitimizes unjust systems rather than transforming them
• Three modes of participation in ML: as work (often uncompensated ghost labor), as consultation (typically performative and short-lived), and as justice (long-term, equitable, rare)
• ML inherits extractive capitalist logic — scale and data extraction conflict structurally with meaningful participation
• Paper "Participation is not a design fix for machine learning" (Sloane, Moss, Awomolo, Forlano, arXiv 2007.02423) grounds the argument
• Recommendations include compensating data labor, making participation context-specific, planning for long-term partnerships, and building a cross-domain failure database
• Amazon Mechanical Turk and reCAPTCHA are cited as examples of unacknowledged participatory labor at scale

Connections: Mit Technology Review · Icml · AI Ethics · Algorithmic Fairness · Participatory Design

Source: https://www.technologyreview.com/2020/08/25/1007589/participation-washing-ai-trends-opinion-machine-learning/