The UK government will need to make big changes if it is to reap the benefits of AI in the public sector, the National Audit Office (NAO) has warned.
While AI is not yet widely used across government, 70% of departments and agencies surveyed by the audit office said they were testing or planning to implement AI tools.
Several departments said they see the technology as an opportunity to automate services, such as application processes, or drafting and summarizing official texts.
However, the NAO warned that the plans are threatened by a range of issues, from unsatisfactory business processes to aging IT infrastructure, risks, skills gaps and data issues.
“AI offers opportunities for government to transform public services and deliver better outcomes for the taxpayer,” said Gareth Davies, director of the NAO.
“To achieve these better outcomes, the government must ensure that its overall program for AI adoption addresses long-standing issues, including data quality and IT aging, as well as incorporating effective risk governance. “.
Almost three quarters (70%) of agencies surveyed told the NAO they were struggling to attract and retain staff with AI skills, with the audit office warning pay levels were too low to attract the necessary staff.
As of October 2022, there were 4,000 digital, data and technology vacancies across government.
Public sector bodies tired of AI risks
Several departments said there was a lack of clarity around legal liability, while they were also concerned about risks of unreliable or inaccurate results from AI, for example due to bias and discrimination, and risks to privacy, data protection or cybersecurity.
They said they needed greater support to address these barriers and share knowledge about how AI is used.
Similarly, controls currently in place for digital and technology spending across government provide no assurance that AI risks have been mitigated, the NAO noted, and the government is still working on its standards, guidance and assurance processes. to support AI adoption and manage risks.
The government needs to establish which department has overall ownership and responsibility for implementing the strategy for AI adoption in the public sector, the report advises, and should establish appropriate roles and responsibilities for those who need to contribute.
The government began work to develop a draft strategy for the adoption of AI in the public sector last year, setting out its ambitions to take advantage of the opportunities that AI presents for public services.
In this year's spring budget, it also announced funding for a number of AI-related productivity initiatives.
However, the NAO said these plans lack realistic measures to address outdated IT systems, address skills gaps, improve data quality and create an AI implementation plan with performance metrics.
“Without swift action to address the barriers preventing the effective use of AI in public services, the government will not realize the benefits it has identified,” Davies said.
The Trades Union Congress (TUC) has also called for greater investment in this area.
The organization said budget cuts over the past decade mean many public services will still struggle to deliver value and improve productivity despite the integration of artificial intelligence tools.
“Fourteen years of budget cuts have left us with old and inadequate equipment. NHS staff tell us it can take 15 minutes to turn on a computer,” the TUC said.
“Imposing new IT systems on these teams without consulting or training staff will be a disaster. We need proper consultation with staff and unions, investment in the workforce and new AI laws to protect workers.”