Behind the label: Glimpses of data labelling labours for AI

ChatGPT is the latest of AI systems to make the headlines for its remarkable computational capabilities. Lesser known and rarely acknowledged is the human labours involved in training and supporting these celebrated AI systems. Thousands of workers, particularly in global south regions, create training datasets, validate model outcomes and mimic computational responses to sustain AI’s research, development and use. Yet little is known about what their work entails. What do data labellers do when they label data for AI?

Drawing on findings from an ethnographic study of data labelling in India, this talk offers insights into the everyday work practices of data labellers, organisational hierarchies, norms, and values that were caught in global flows of resources, rhetoric, and relations of power. We trace these practices, norms and frictions to better understand their influences on everyday annotation work as well as answer an important question, why should we, AI researchers and practitioners, concern ourselves with these seemingly distant realities?

Presentation slides
Speaker Details

Srravya is a doctoral researcher at the Centre for Human Computer Interaction Design (HCID) at City, University of London. Her PhD work examines data annotation work practices, paying particular attention to systemic challenges and frictions, to envision and inform just, equitable futures of AI. More broadly, her research contributes to socio-technical studies of emerging technologies, in which she strives to adopt and develop a power-aware perspective.

Date:
Speakers:
Srravya Chandhiramowuli
Affiliation:
City, University of London

Watch Next