Using artificial intelligence to automate and redirect calls could save the DWP and HMRC millions of hours each year, according to a new report that calls for the science department to be given greater powers to advise departments on the use of AI and automation.
The Social Market Foundation has called for the Department for Science, Innovation and Technology to act as a “one-stop shop” for all public sector AI and automation needs, supporting other departments to identify areas where they could save resources by adopting these approaches and issuing guidance.
To illustrate the potential of AI and automation for government, the cross-party think tank said two of government’s biggest departments could drastically cut down on call times by automating and “deflecting” calls from citizens.
Citizens spent the equivalent 6,895 working years on the phone to the Department of Work and Pensions and HM Revenue and Customs in 2022-23. Today’s report estimates that two-thirds of these calls could be automated, based on case studies where automation and AI tools were introduced in similar sectors – saving 8.12 million hours, or more than 4,300 working years.
Other use cases in the report include cutting down on paper-based decision-making at the Driving and Vehicle Licensing Authority, where more than 200,000 DVLA medical-licensing decisions – nearly 20% of the total – took over 90 days in 2023.
“While estimating time savings is difficult, a majority of these decisions are paper-based, meaning AI and automation have significant potential to decrease user wait times through technologies such as OCR and triaging,” the report said. OCR, or optical character recognition, could be used to categorise and extract documents internally, while AI could help to prioritise cases by identifying the most complex applications.
But the report found a number of “structural impediments” to adopting automation and AI in the civil service, including aversion, difficulty scaling pilots, and a lack of institutional capacity or expertise.
Other barriers include a reluctance to fund training or AI projects that could provide long-term savings; and siloed decision-making, “with vertical organisational structures that do not maximise horizontal coherence across the public sector when it comes to AI/automation adoption”.
DSIT should be 'empowered' on AI and automation
The report calls for the team in DSIT’s so-called "digital centre of government" to be “empowered” to advise departments on promising areas of opportunity – and “push back on automation proposals from departments that are too unambitious, unrealistic, or do not have proper regard to interoperability”.
The department should appoint a head of citizen experience – at director level or higher – to oversee departments’ automation strategies and identify opportunities to save citizens’ time, such as by removing duplication or forms, it says.
And it also calls for the department to play a central role in training civil servants across government by setting up a “data academy” focused on AI and automation skills.
DSIT should also be tasked with synthesising evidence on evaluations of AI and automation that is already being used across the UK government or in similar jurisdictions and make this available as a central resource to leaders in government departments and public bodies, the report says. The department should also provide guidance on best practice for evaluating AI and automation interventions, including approaches to benchmarking and monitoring, it adds.
The department should use its oversight of AI and automation across government to shortlist use cases with the most extensive evidence base and the strongest potential to significantly improve performance gains, according to the think tank.
And it should be given control of a “transformation fund” that it can use to disburse projects it has shortlisted. Should this not happen, it says the department should work with the Treasury and the National Audit Office to identify ways to fast-track shortlisted projects.
Cross-government culture change
The report also calls for broader changes across departments to make it easier for them to adopt new technologies that could save time and money.
It says that each department should appoint automation leads, reporting to the citizen experience head, to identify opportunities and monitor new automation projects.
Across government, the report calls for the business case process for spending on AI and automation to be “streamlined”, with fewer, shorter business cases being submitted. These processes “should aim to strengthen a culture of ‘invest to save’”, the report says, with in-year savings made to departments’ day-to-day budgets used to fund capital investment.
DSIT could aid these efforts by sharing model business cases and consulting with departments on building their business cases, the report suggests.
The think tank also calls on departments to adopt a “parity of risk principle” that requires departments to consider the risks not just of adopting new AI or automation processes, but also the corresponding risks of not doing so, in risk assessments and internal processes.