Controversies in the Data Society seminar series 2026

Srravya Chandhiramowuli is a Post Doctoral Research Fellow and Thematic Lead on Data Work in the Planetary AI project, at the University of Edinburgh. Her research closely follows the on-ground practices of dataset production for AI, bringing particular attention to systemic challenges and frictions in data work and AI supply chains. Building on scholarship in Human Computer Interaction (HCI) and Science and Technology Studies (STS), Srravya’s research seeks to contribute towards just and equitable futures in technology

Beyond the label: repairing data work, rethinking AI  

Despite recent advances in AI’s computational capabilities, data work—the human labour required for training, fine-tuning, and evaluating AI systems—remains indispensable to AI production. Yet, data work is constituted as a routine and repetitive activity, with little scope for applying expertise and skill, and often conducted under unfair conditions of work. As the restrictive, uneven structures shaping data work become increasingly visible, there is a crucial need to consider how data work might be repaired and reoriented towards a more just and equitable practice. In this talk, I will present reflections on some of the work that lies ahead in this regard, drawing on my ethnographic engagement with a civic-tech initiative based in India that builds datasets for training and evaluating online safety systems. Adopting a feminist orientation, they produce safety datasets by collaborating with those most impacted by online harms, inviting them to contribute and annotate data on online harms. Drawing on insights from two dataset projects developed in this orientation, I highlight how this approach reorients data work as a site for repair and redress by enabling a wider scope for contributions beyond discrete tasks and recognising contributors as experts. This recognition then surfaces limits and tensions in advancing just reward for data work, and contributors’ role in governing the datasets they help produce. Thinking with this case, I will share insights into the open challenges in translating alternative, feminist visions for data/AI into actual practice.

This video by Prof. Jeff Collin of the University of Edinburgh is available under a Creative Commons Attribution Non Commercial No Derivatives 4.0 licence.