Government plans to use artificial intelligence (AI) to crack down on benefit fraud are facing questions over fears bots could make biased decisions.
Though the government suggests AI can prevent overpayment of benefits by assessing historical data, campaigners argue that the machine learning models could make biased decisions which unfairly affect vulnerable people.
Groups have also scrutinised the government’s unwillingness to share details of the plan, suggesting “well-documented serious risks” in AI-informed decisions were being ignored.
“Expanding the technology while ignoring calls for transparency and rigorous monitoring of and protections against bias will risk serious harm to vulnerable families,” Child Poverty Action Group said.
However, the government has previously declined to publish details of the AI trials over fears that fraudsters could learn how to manipulate the new technology.
Some £8.3bn is anticipated to have been overpaid in benefits last year, prompting the government to announce a £70mln funding boost to the trials earlier this month.
This could help save up to £1.6bn from fraudsters targeting the welfare state over the next two years, according to The Department of Work and Pensions.
Concerns of unintended bias and misinformation still remain though, campaign group Privacy International has warned, leaving questions on the technology’s use unanswered.
A spokesperson for the DWP defended the move meanwhile, explaining: “AI does not replace human judgment when investigating fraud and error.
“We continue to explore the potential of new technologies in combating fraud and deploy comprehensive safeguards when doing so.”