All Things Policy Invisible Data Workers Powering Machine Learning Models
AI systems depend on vast amounts of labelled data, produced by “data workers” who tag images, categorise text, and review content so machine-learning models can function accurately and safely. While the data-annotation market is rapidly expanding, the labour behind it often remains invisible. Much of this work uses BPO firms in countries like India, where workers - frequently women - are recruited with promises of flexible work-from-home jobs. In reality, many face vague contracts and limited legal or mental-health protections. As AI scales globally, this raises urgent questions about the ethical sustainability of data annotation and who should be responsible for safeguarding the workers who power it. Join Kripa Koshy in conversation with Bharath Reddy and Suman Joshi as they unpack all this and more in this episode.
All Things Policy is a daily podcast on public policy brought to you by the Takshashila Institution, Bengaluru.Find out more on our research and other work here: https://takshashila.org.in/research-areas