How government departments can unlock the potential for AI

By on 11/12/2023 | Updated on 11/12/2023
A picture of attendees at Global Government Forum's roundtable with UIPath
Global Government Forum Roundtable with UiPath.

There is real interest in the use of artificial intelligence in government among public servants, but there is not yet an understanding of how it can be fully utilised and deployed, discussion at a Global Government Forum roundtable has revealed

GGF and knowledge partner UiPath brought together senior civil servants to discuss the potential for AI-powered automation to meet the efficiency challenge faced by government.

The UK government, like others around the world, faces a daunting workload in the years ahead. The chancellor Jeremy Hunt has launched a public sector efficiency drive, with the aim of boosting productivity to match the projected annual increase in demand for public services.

The chancellor named AI as one of the ways this can be achieved, and this session brought together civil servants from across government to discuss the potential for AI with found strong interest in deploying it among those attending.

The session also identified potential barriers to its use including a lack of clarity on potential use cases and structural problems blocking deployment in government.

The opportunity of AI

The roundtable was held under the Chatham House Rule – meaning that we won’t identify those speaking. Attendees at the session all thought there was potential for AI to be deployed in government to help address efficiency challenges.

“I think we collectively have a bit of an opportunity at this point,” said one attendee, while another added that AI was a “huge opportunity for streamlining how we offer a much more modern service to people within an extremely complex regulatory environment”.

Attendees also said there was increasing interest in AI from departmental permanent secretaries.

“From a senior [official’s] perspective, they see AI as a sort of a magic wand to solve government problems – to improve efficiency, productivity, create savings, modernised government, and they think it can be done very quickly,” said one attendee.

Submissions to the chancellor’s productivity review had added to this interest among ministers in the use of AI, which an attendee revealed had led to the creation of a cross-public sector group tasked with looking at strategy, vision, implementation, and adoption of AI in government.

Addressing the ‘strategic imperative’ to develop AI

It was clear from the discussion that, in the words of one attendee, better use of AI is a “strategic imperative” to the work of government.

“It will be impossible to do what we want to do without significantly increasing the amount of automation that happens across the department,” they highlighted.

Much of the discussion at the session covered how government could drive this development. The same attendee said they were looking at what they called “safe acceleration” of the use of AI, looking at launching five “proof of concept” projects to help develop the governance, legal and operational provisions around the use of AI.

Addressing the efficiency challenge faced by government is very important, they said, but it is also vital to acknowledge the potential in terms of the customer experience. “They’re not mutually exclusive by any means, but they often are treated as such. If we look slightly beyond the incredibly important issue with efficiency we can see what these technologies can do for changing the way that we relate to citizens, and changing the types of services that we offer.”

Sharing AI knowledge across government

Another key element of the discussion was how government departments could share information on how they are using AI with many attendees keen to learn from other departments rather than duplicating existing areas of work.

One senior departmental figure mentioned their own trials with generative AI, using freely available tools.

These include using AI as part of ministerial correspondence – an area where attendees said there should be collaboration between departments to avoid duplication.

“Every government department has to do the same kind of thing, so it would definitely be interesting to share what other people are doing on those kinds of topics,” said the official whose department is working on the pilot.

Another senior figure agreed ministerial correspondence was an example of an area where government needed to get a better understanding of the work that was ongoing across departments.

“I see all this great work that’s going on in departments, but it is going on in departments, as multiple people have touched on [rather than being coordinated centrally]. In correspondence management each department is going to develop their own solution because that is imperative to the efficiency – unless someone drives a different route.”

Government is not “in a particularly good place, being very honest, on how we work collaboratively across government on solving some of these solutions”, they said, although they added that it was working to boost the capacity to drive coordination.

“I think there are areas within each department that are doing fantastic stuff. But I think we’re missing out on that opportunity to aggregate all of that work to create leaders [in certain services] – departments that are going to lead on correspondence management and create a solution that is scalable to the rest of government.”

Such an approach would also require providing capability to these trailblazing departments, including from industry partners such as UiPath, attendees said.

“Can we harness some of our industry partners to create capability for departments to draw on, to deliver on specific use cases?” one asked.

Philip Sheen, the head of public sector, UK & Ireland for UiPath, offered to help catalyse these conversations. “We can make available the use cases of areas that are common between all of you, and others outside this room, across where AI and automation is being used in government.”

Global Government Forum Roundtable with UiPath

Tackling legacy technology and ‘operational debt’

As well as sharing existing use cases, the need for greater understanding of where and how AI could be usefully deployed was a recurring theme during the discussion.

In one department that had used AI in pilot projects, the use cases had been developed by officials on the frontline who indicated areas where they thought AI and automation could help.

However, several attendees highlighted there was a lack of understanding about where such technology could be used.

“I think the challenge is that a lot of our people don’t know what they don’t know – they have no concept necessarily of what technology could do for them,” one said.

As a result, their department had started working in an informal way to identify use cases. “But we need a bit more activity in generating specific use cases, and to do that you need to understand what technology does and try and match them together. It’s almost chicken and egg here, because we’re not an intelligent client in that sense in terms of understanding what we could ask for.”

“Chicken and egg completely resonates,” said another, who highlighted another problem – what they called the “operational debt” of running government on old systems and processes.

Solving some of the underlying problems caused by these legacy operating systems for government was more important than, for example, possibly using AI to quicken document reading, they said.

“It’d distil that information quicker. That’s great [but] is it worth rolling that out when the substance is still on our creaky systems? It will be better to invest and build something that has much better capability [overall], rather than building a sophisticated layer on top of what we’re doing.”

This requires a much broader conversation than simply discussing technology. “How do you get out of that cycle and make that case for change?” they asked.

New ways of working needed

Breaking this cycle is where culture change is required, attendees agreed – changing what one senior official who had also worked in the private sector called the “very hierarchical, functional setup” of government departments.

“In my experience, this is probably where government departments are further behind than commercial organisations,” they said. “It’s not so much the use of technology or availability of money or resources, it’s more the structures that have been ingrained and may be decades old.”

Among the solutions proposed by officials to this problem was creating multidisciplinary teams to better join up the needs of the frontline with available technologies.

“I don’t think there’s an appetite within our organisation to throw everything up in the air and redesign organisational structures around outcomes, so we have to do what can we do within those boundaries,” said one official.

Another official spoke of an innovation they were working on to improve datasets in their department, and agreed they had faced more cultural, and people challenges than technical ones.

“The multidisciplinary working piece is something we have really struggled with,” they admitted. “When you have a school of thought from DDaT [the digital, data and technology profession] on how you should develop things, from the analysis function, and the policy profession, it can be challenging, certainly. But the intention is trying to change the way we do things.”

Their lessons from this process had been around “really thinking through” the processes to deliver services to their department. “We’ve reorganised several times along the way to try and invest a lot more in things like commissioning and engagement to get good requests [of work] to do.”

This comes back to the need to talk to different groups across departments about their requirements – and how AI can be used to meet them.

“What they originally asked for is, almost always, not what they get. But [we’re] having that discussion around what we can deliver and working with them as experts. It’s been something that’s really become incredibly useful, and I defend very strongly, because I think it’s the only way, we can understand what the business needs.”

The How your department can use AI-powered automation to meet the efficiency challenge roundtable was held on Wednesday 18 October in London, and was attended by:
Chris Thorn, Chief Digital Officer, Department for Environment, Food and Rural Affairs
Elinor Godfrey, Deputy Director and Head of Centre for Data & Analysis, Foreign, Commonwealth & Development Office
Giles Hartwright, Deputy Director, Modernising Technology, Central Digital and Data Office
Guy Gibbs, Head of FCDOx & Head of Strategy and Futures, Foreign, Commonwealth & Development Office
Joanna Fiddian, Deputy Director and Head, Transformation – Legal Aid Agency, Ministry of Justice
John Kelleher, Area Vice President, UK & Ireland, UiPath
Philip Sheen, Vice President for Public Sector UK & Ireland, UiPath
Rachael Beattie, Assistant Director, Transformation Directorate, HMRC
Rose Macfarlane, Deputy Director, Delivery and Strategy, Department for Environment, Food and Rural Affairs
Simon King, Head of Digital Strategy, Innovation and Business Architecture, Department for Work and Pensions

About Partner Content

This content is brought to you by a Global Government Forum, Knowledge Partner.

Leave a Reply

Your email address will not be published. Required fields are marked *