In Telangana, an algorithm, initially developed to profile criminals, now decides who is poor. And it has wrongly identified thousands as not worthy of welfare benefits due to faulty data and bad decisions.
The result is disastrous. They are cut off from receiving welfare benefits, including subsidised food that the government must provide to the poor under the Indian food security law.
Since 2016, Telangana has been using Samagra Vedika, an algorithm developed by a private company to check citizens' data in several government databases to create comprehensive digital profiles or “360-degree views'' of its 30 million residents. The system has been deployed to eliminate welfare frauds and determine whether an individual deserves government doles.
Over the past year, The Collective’s member, Tapasya, along with Kumar Sambhav and Divij Joshi investigated the use and impact of such welfare algorithms with support from the Pulitzer Centre’s Artificial Intelligence Accountability Network.
The first part of the series reveals how unfettered use of Telangana’s profiling software deprived thousands of poor people of their rightful subsidised food for years. Its secret codes, judging people, have been playing god with the lives of the poor, with apathetic officials often refusing the necessary human oversight in rectifying the machine error.