Artificial intelligence (AI) could help automate a vast range of the work done by civil servants across hundreds of government services, research suggests.
Generative AI (GenAI) could extend that even further, but unions and experts have warned that the introduction of new technology must be done in a way that benefits everyone – and not just as a chance to cut jobs.
According to the researchers at the Turing Institute, the government carries out around one billion citizen-facing transactions per year, across almost 400 services.
The researchers looked at 201 services that involved a decision and an exchange of information with government, such as registering to vote or applying for a national insurance number. These are the ones that consume the most effort within government and have the highest potential for time saving if they can be automated.
These services were made up of around 143 million complex but repetitive transactions of which 84% could be easily automated, “representing a huge potential opportunity”, the researchers said.
“AI has enormous potential to help governments become more responsive, efficient and fair,” said Jonathan Bright, head of AI for public services at The Alan Turing Institute. “Even if AI could save one minute per transaction, that would be the equivalent of hundreds of thousands of hours of labour saved each year. Achieving responsible and accurate automation with AI will require a lot of work, however the huge benefit justifies the investment needed.”
The research paper AI for bureaucratic productivity noted that there is “considerable excitement” within government about the potential of AI to improve public service productivity by automating complex but repetitive tasks.
Delivering many government services can be complex and often involves a chain of administrative and bureaucratic decision-making. That might mean civil servants analysing application documents, confirming identities or making judgments on eligibility. This work is necessary but these long chains create delays and run up costs for the government and the public.
“Especially with the rise of generative AI as a product ready to be integrated into organisations, there is currently optimism around the idea that government services can be made faster and more citizen-oriented, both improving public satisfaction, and cutting costs and bureaucratic overhead,” the report said.
While the 201 decision-based services identified by the researchers only account for 15% or 143 million of the total 965 million of yearly government transactions, they are some of the most time consuming which could mean a large payoff if they were automated.
For example, some of the most common tasks included recording, preparing, sorting, classifying and filing information, verifying the accuracy of information provided, and filing and storing completed documents on computer hard drive or disk.
While there was variation between government departments, many services were above the threshold that would make them good candidates for automation. The researchers said all 20 services from the DVLA as well as the DVSA are automatable, but there were far fewer services at the courts service or within the Department for Education or the Department for Work and Pensions that could be automated.
Topics such as driving and transport as well as education, training and skills appeared to be the most automatable, while the topics of benefits, childcare and parenting, and national security appeared to be the least automatable.
However, the report noted: “We would emphasise that we do not believe the labour of these professionals is directly replacable by AI – and very few occupational categories can or should be entirely replaced by technology, especially in public services.”
But it said the technology could significantly enhance individual and collective departmental productivity by streamlining bureaucratic tasks, and thus freeing up time for workers to focus on tasks requiring human judgement, creativity, discretion, and decision-making.
But that’s not all – the rise of GenAI may mean that even more of the public sector could be automated.
The researchers also looked at 20 of the 201 decision-based services that had a low percentage of routine tasks within them. Out of the 20 services, 14 had medium potential to use generative AI. These services included apply for benefits and visas, as well some services for businesses and self-employed individuals.
“Hence, according to these results, the arrival of GenAI shows even higher potential for automation than might be expected solely by analysing the proportion of routine tasks,” the researchers said.
Governments have been looking to technology to increase the productivity of the civil service – and reduce the size of it – for decades, with varying degrees of success. However, AI does seem to be offering up some new opportunities – Derby Council recently said it hopes to save tens of millions of pounds by using generative AI across its services in the next few years.
Dan Lucy, director of HR research and consulting at the Institute for Employment Studies, said AI typically affects tasks rather than jobs, and driving efficiencies can release staff to focus on higher value-added activities.
At its simplest that could mean shifting from answering routine queries to better addressing more complex ones. But it can also open up the possibility of doing different things and offering new services, both of which may lead to new jobs requiring different skills, he said.
“For example, the National Grid now uses drones rather than people to inspect power lines, but this then requires expertise in operating drones effectively,” he said. “Adoption and implementation of AI can have substantial benefits, but only if aligned to the development of a strategic plan for the workforce which enables those benefits to be seen and any risks to be effectively mitigated.”
And Jon Richards, assistant general secretary at public service union Unison, said that while new technologies have the potential to transform the delivery of public services, that requires their careful and responsible introduction and use.
He said health, education, police, care and local government organisations are all trying to get by and keep services running with insufficient resources and too few staff. If AI can be used to run repetitive, bureaucratic tasks more efficiently, that might free up staff to work on the more public-facing elements.
“But it’s not without risks too. Neither the government nor public service employers should be eyeing up AI just to save money by making yet more cuts and redundancies. Unions, employers and ministers must work together to ensure AI is used in a way that benefits everyone,” he said.