Approximately four-fifths of government decision-based services are highly automatable, according to new research.
According to the Alan Turing Institute report, this figure includes all 20 such services provided by the Driver and Vehicle Licensing Agency and the Driver and Vehicle Standards Agency. DVSA is already using AI to detect suspicious activity in MOT applications.
Researchers at the government-funded institute found that central government departments and other public bodies in the UK carry out around 1 billion 'citizen transactions' a year across 377 services, such as passport applications and Universal Credit processing. It was determined that there was.
They focused on 201 of these services, focusing on services that “involve decision-making and the exchange of information between government and the people”, such as registering to vote and applying for a National Insurance number.
According to the institute, these services are made up of approximately 143 million complex but repetitive transactions that have high potential for automation with AI. The researchers added that these “are the most labor-intensive within government and have the greatest potential for time savings if automated.”
The study found that 84% of services can be easily automated. This is defined as 75% of transactions being “fully routine” tasks and 69% of services being defined as consisting of 90-100% “fully routine” tasks .
Among the government services analyzed by the institute, the service most difficult to automate is “appeals of visa or immigration decisions,” with approximately 38% of tasks related to this provision being routine. This was revealed through an investigation.
Such services “may be expected to necessarily retain a significant human element,” the institute said.
The study also grouped services into 'topics', with areas that could be most automated such as 'driving and transport' and 'education, training and skills', while topics such as 'benefits' and 'parenting and ”, “National Security”, etc. The least automatable.
This study focuses on service transactions as a “bundle of tasks rather than a person assisting in executing the transaction.” The institute said this focus “allows us to explore the potential of applying AI and machine learning to specific aspects of service delivery, without assuming that entire occupational roles need to be automated.” .
Dr Jonathan Bright, Head of AI for Public Services and Head of Online Safety at the Alan Turing Institute, said: “AI has huge potential to help make governments more responsive, efficient, and fair. Even if AI saves him one minute per transaction, that’s hundreds of thousands of hours a year. This equates to savings in labor time.
“Achieving responsible and accurate automation with AI requires a lot of work, but the benefits are significant and justify the investment required.”
However, the study acknowledges some limitations.
- It only takes into account the bureaucratic transactions that initiate access to the service, not the transactions required to continue the service or maintain access to the service.
- This work was primarily done through desk research without collaboration with the service providers themselves. Therefore, it is not possible to verify whether the mapping is complete.
- There is frequent debate about how easily tasks can be classified as routine or non-routine, and how this definition varies across contexts and professions.
- In the absence of clear time use data, researchers assume that the same amount of time is spent on each task, which may result in overestimating or underestimating the potential for automation.
“For these reasons, further work within government that attempts more detailed task-based mapping, especially work that involves direct involvement, would be of great value.”
By collaborating with service delivery professionals, we now have greater visibility into the nature of day-to-day operations,” the report states.
In addition to automating routine tasks, this study also looked at how non-routine tasks can be automated through generative AI. They found that approximately 70% of services with a low proportion of routine tasks have a “moderate chance” of being automated by this emerging technology.
Last month, Cabinet Office Secretary Oliver Dowden announced plans to invest £110m in the Incubator for AI, the Cabinet Office's AI arm, and emerging technologies, stating that AI is “the only way to reduce the size of the civil service.” “It is a sustainable path.” It's to speed up what he described as a “dog job.”
A government spokesperson said: “Artificial intelligence has the potential to revolutionize public services and improve productivity. We welcome the findings of this study and look forward to new ways of leveraging AI, including further automation of government services. We are continually exploring.
“For example, the government has committed to converting at least 50 of the top 75 government digital services to a 'good' standard by next year, making them easier to use, faster and more efficient, including responsible use. Make available. About automation and AI.
“We have never seen AI simply as a tool to replace jobs. Instead, by eliminating time-wasting administrators and bureaucracy, we can free up civil servants to do their real work. So we can solve problems and serve our people.”