**Fears rise in benefits system automation could plunge claimants deeper into poverty
DWP spending millions on " Intelligent automation garage " to develop welfare robots to replace humans.**
( " Zero hour contract … does not compute … does not compute … does not … " ??? )
**_The UK government is accelerating the development of robots in the benefits system in a digitisation drive that vulnerable claimants fear could plunge them further into hunger and debt, the Guardian has learned.
The Department for Work and Pensions has hired nearly 1,000 new IT staff in the past 18 months, and has increased spending to about £8m a year on a specialist “intelligent automation garage” where computer scientists are developing over 100 welfare robots, deep learning and intelligent automation for use in the welfare system.
As well as contracts with the outsourcing multinationals IBM, Tata Consultancy and CapGemini, it is also working with UiPath, a New York-based firm co-founded by Daniel Dines, the world’s first “bot billionaire” who last month said: “I want a robot for every person.” His software, used by Walmart and Toyota, is now being deployed in a bid to introduce machine learning into checking benefit claims.
The DWP is also testing artificial intelligence to judge the likelihood that citizens’ claims about their childcare and housing costs are true when they apply for benefits.
It has deployed 16 bots to communicate with claimants and help process claims and is building a “virtual workforce” to take over some of the jobs of humans. One recent tender document requested help to build “systems that … can autonomously carry out tasks without human intervention”.
The developments emerged during a Guardian investigation into one of the most radical but least understood welfare reforms since the roll-out of universal credit that will apply to 7 million people.
The DWP believes welfare transactions could be handled more quickly, accurately and cheaply using robotic process automation and is developing it for use in UC.
But claimants have warned the existing automation in UC’s “digital by default” system has already driven some to hunger, breakdown and even attempted suicide. One described the online process as a “Kafka-like carousel”, another as “hostile” and yet another as a “form of torture”. Several said civil servants already appeared to be ruled by computer algorithms, unable to contradict their verdicts.
“We are striking the right balance between having a compassionate safety net on which we spend £95bn, and creating a digital service that suits the way most people use technology,” said a DWP spokesperson. “Automation means we are improving accuracy, speeding up our service and freeing up colleagues’ time so they can support the people who need it most.”
But Frank Field, chairman of the Commons work and pensions select committee, warned that vulnerable claimants “will be left at the mercy of online systems that, even now, leave all too many people teetering on the brink of destitution”.
“We’ve already seen, in the gig economy, how workers are managed and sacked, not by people, but by algorithms,” he said. “Now the welfare state looks set to follow suit, with the ‘social’ human element being stripped away from ‘social security’.”
Key details about the automation push remain secret. The DWP has refused freedom of information requests to explain how it gathers data on citizens. Simon McKinnon, the chief digital and information officer of DWP Digital, said this year it was developing a way to “build a holistic understanding of digital personas”, but refused to say what information was gathered to do this.
The ministry has previously told parliament it gathers data from private credit reference agencies, the police, the Valuation Office Agency, the Land Registry and the National Fraud Initiative, which gather information from public and private bodies. But it is now declining to update the list, claiming it would “compromise the usefulness of that data”.
“There are concerns that government is accelerating the automation of the welfare system without a proper evidence-based consultation about its impacts,” said Dr Lina Dencik, co-founder of the Data Justice Lab at Cardiff University.
“Users have already raised significant problems and we need to understand which groups are more likely to be targeted by algorithmic decision-making than others to understand if it shows bias.”
The centre of the robotics drive is the DWP’s “intelligent automation garage”, a unit based in Newcastle and Manchester where 32 programmers design, build and run automations.
Staff are using UiPath to develop machine learning to check claims for fraud, which suggests welfare computers will autonomously learn and alter the way they make decisions with minimum human intervention.
One recent staff member at Newcastle told the Guardian they already “have ways of creating a digital image of somebody”. He stressed much of the work was secret, but said this did not mean it was against citizens’ interests.
The digital transformation is costing hundreds of millions of pounds. The DWP Digital’s budget has risen 17% to £1.1bn in the past year and IT firms have been awarded huge contracts to help run the system. The DWP is also rapidly expanding its own private technology company Benefits and Pensions Digital Technology Services, which recruited more than 400 staff in the year to April, while DWP Digital recruited 520.
A spokesperson for the DWP insisted it was using AI to help people find work, to reduce the burden on claimants to prove their circumstances, and to help vulnerable people access welfare more easily without having to provide evidence of their digital identity.
“We want to be supporting staff and citizens with the information they need, at the right time, and not disadvantaging those who are vulnerable,” they said.
However, there is evidence of rising error rates in parts of the welfare system that have already been automated. A system of realtime data-sharing between the HMRC tax office and the DWP about universal credit claimants’ earnings is triggering more and more disputes, with the rate rising fourfold between May 2017 and October 2018, according to the government’s own figures, with up to 5,700 people a month affected._**