There is not much evidence around about the effectiveness of artificial intelligence (AI) in public services. So, the renowned technology thinktank the Alan Turing Institute had a go at providing some earlier this year.
According to a paper published by its researchers, AI could help automate about 84 per cent of repetitive transactions across 200 government services.
In a study released to co-incide with its AI UK conference, the institute’s analysis showed that the UK government carries out about a billion citizen-facing transactions per year spanning almost 400 services – including passport applications and universal credit processes – and 57 departments.
Researchers focused on 201 of those services that involved a decision and an exchange of information between government and a citizen, such as registering to vote or applying for a national insurance number – which were identified as services that consume the most effort within government and therefore have the highest potential for time saving if they can be automated.
They went on to estimate that these services were made up of about 143 million complex but repetitive transactions, giving them a high potential for automation by AI. And that’s how they reached the 84 per cent figure for being ripe for AI pickings.
“Automation of repetitive bureaucratic tasks is where I think the big opportunity area lies,” says Dr Jonathan Bright, head of AI for public services and head of online safety at the Alan Turing Institute.
“An enormous amount of public sector time is spent doing repetitive things such as filling in forms or assessing eligibility which could either be assisted by AI, or perhaps to some extent slightly replaced by AI in a way that would free up time for other activities.”
One of the most striking examples of that is adult social care. Bright explains that assessments of adult social care needs are usually conducted by visit and interview. Those notes then need to be written up in a statutory report, which is a time-consuming process.
The technology already exists to record and transcribe such interactions, but apps are now offering AI capabilities which can reformat those conversations into statutory documents.
“This just saves hours of time,” adds Bright. “And you can find this across education, across child social care, across the emergency services, all sorts of frontline public sector work where professionals typically respond really well to these technologies when they’re placed in their hands.”
Adoption of such technology could reduce the burden of bureaucracy significantly, allowing those workers to spend more time building better relationships with the people they are working with, says Bright.
Those examples of the use of AI bring more than just productivity benefits. If the technology can be harnessed to remove the more mundane aspects of work, it can improve job satisfaction.
“There is a lot of burn-out,” adds Bright, pictured below. “You find this across education, healthcare, policing, prisons, asylum, you name it. And I think that’s where there has been real interest in AI over the last year and a half.”
Trying to make the case for AI adoption within government remains quite a challenge. There are legitimate concerns around outsourcing human decision-making to machines, which can be no better illustrated than the Post Office Horizon scandal. With the advent of AI, those concerns are arguably put into even sharper focus.
However, it must be stressed that government officials are putting a lot of thought into how they get it right when it comes to ethics, privacy, transparency, skills, safety and security.
The UK Government document, “A guide to using artificial intelligence in the public sector”, stresses that AI is already being deployed in a number of public sector organisations, from fraud detection to answering customer queries.
National Grid, for example, has turned to AI to help it maintain the wires and pylons that transmit electricity from power stations to homes and businesses. And the GovTech Catalyst was set up to help public sector bodies take advantage of emerging technologies.
The UK Government also set up three new bodies to support the use of AI, build the right infrastructure and facilitate public and private sector adoption of the technology.
The AI Council, the Office for AI, and the Centre for Data Ethics and Innovation are all working towards mainstreaming the technology in multiple sectors, including government.
But its adoption is not necessarily a panacea for productivity. Bright adds: “It’s very striking to see hardly any increase, or arguably no increase at all, in public sector productivity over the last 20 years, and that’s even with the introduction of a lot of digital technology into government.”
Bright says there are a variety of explanations for that including that it is very hard to measure public sector productivity, but he highlights a particular concern.
“One of my biggest fears about AI is that we’ll squander a lot of the savings by increasing requirements,” he adds. “So, let’s say it’s now really quick for me to write 1,000-word statutory reports, and as a result my manager wants a 5,000-word statutory report. You have to be careful that you don’t end up spending the same amount of time doing a report and bureaucracy as you did before.”
Nevertheless, he remains “bullishly pro-automation”.
“Not for everything, as there are plenty of complicated decisions and situations which need human attention, but there are also plenty of decisions that don’t get enough human attention now because humans are overwhelmed with decision-making tasks,” Bright says.
“And also we have a poor understanding how often these sort of processes fail now, so if you bring in AI you will want to measure how often they are making decisions correctly, and compare it to the status quo ante.”
On the skills front, there’s also never been a better time for the adoption of AI.
“I’ve been studying government digitisation, largely in the UK, for more than 10 years, and I’m not exaggerating when I say I’ve never been more optimistic than I am now,” adds Bright.
“That doesn’t mean it’s going to be really easy to do at the drop of a hat, but I think in the last 10 to 15 years there’s been a growing consciousness that having technology expertise in-house in government – compared to before where the focus was almost entirely on outsourcing – is really important,” he adds.
Another plus is that with generative AI, you don’t necessarily need programming skills.
“If five years ago you’d asked me to make a lesson plan tool for teachers, I’d have said, ‘Well, this is something you need specialists to develop,” Bright says.
“Now a teacher can open up a Gen AI app and ask it to make a lesson plan. That doesn’t mean you don’t need specialist knowledge but the interaction between the technology and the person is much closer.”
He references a quote last year he really liked from the renowned AI developer Andrej Karpathy who said: “The hottest new programming language is English.”
Public acceptance is also growing. Bright says it depends on the context in which AI is used, and some areas are inherently riskier than others – for example, driverless cars and in medical environments. But even autonomous vehicles are gaining traction, and he thinks they will eventually come to be seen as “standard”.
But what about AI hype and, equally, doom-mongering about generative AI being an existential threat?
“There was undoubtedly a hype cycle we entered into in 2022 when people first got to grips with the fact that this new technology exists.
“There were some who thought we were only months away from artificial general intelligence. But the reality is that people are making products out of this technology now, and although it doesn’t solve everything there is a lot of potential there.
“What’s struck me is how enthusiastic people have been about the technology, which is very unusual for people in the public sector who are used to having technology imposed on them from the top down, and people being frustrated with technology all the time.
“So, it’s striking the positivity with which the technology was received, and I’m talking about generative AI, in particular.
“And part of the reason for that is when people start to perceive technology as helping them with their day-to-day work, rather than imposing another set of constraints and obligations on them. That was very striking, and I think a cause for cautious optimism.”