Local authorities are on the cusp of a digital revolution, especially with AI’s ability to transform public services. But there are still many avenues that need to be explored, and many questions that need to be answered.

At LGPN North last year, our panellists explored how AI is transforming local government, discussing everything from training and adoption challenges to risk management and ethical considerations.

Here are the key takeaways from the conversation…

Transforming and empowering local government

When it comes to transformation, local government has always been a little slow to respond. But with the rapid advances in AI, local government finally has the chance to drive meaningful change when it comes to public service. However, to drive change, local authorities need to win the support of its staff and the public when it comes to using AI. Kicking off the discussion, our panel host posed the question, ‘What does AI empowerment mean to you?’

Timothy Spiers, Director, IT Innovation, Digital Transformation at Oxfordshire County Council acknowledged the hype around AI but also emphasised the need for risk management and responsible implementation. ‘It is an opportunity, and I think if we get the risks right, if we manage that, then we can really make those improvements, which let’s all be honest, we’ve been trying to make for many years.’ He continued, ‘I think AI has the opportunity to really boost us. I think it is about harnessing the power of AI in a responsible way.’

Madeline Hoskin, Assistant Director Technology, North Yorkshire Council spoke getting used to AI and experimenting with its different uses. ‘For me at the minute, that means really leaning into being experimental, really leaning into moving forward with small bits of AI testing, having a low confidence but leaping anyway, being brave and building those user cases across those different services.’

Daniel Dredge, Programme Manager, Commissioning and Transformation, Spelthorne Borough Council offered a practical perspective, describing AI as a tool to enhance efficiency and service delivery. ‘I think the impairment with AI is just meeting our customer needs better. So doing what we do now but better. It’s a tool to create and improve digital services, be more efficient, be more productive.’ He continued, ‘I find it saves a lot of time. I just think it makes things more accessible for people getting the help they need faster. But I don’t want to lose sight of the fact. I just think that you should design a good digital service. That should be the goal and AI’s just a tool to help you get there.’

AI training and adoption challenges

But to truly benefit from AI, staff need more than just access to the tools – they need training, particularly when it comes to creating effective prompts and understanding the limitations of AI systems. As Timothy Spiers highlighted, while tools like Microsoft Copilot may seem intuitive and ready ‘out of the box,’ there’s a lot more that goes into using AI than meets the eye.

One concern Timothy identified was AI generating incorrect or misleading information, known as ‘hallucinations.’ These occur when AI generates inaccurate or made-up responses and introduces them as fact. As he put it, ‘One of the things I found out recently in terms of the risks, is this thing called hallucinations in AI.’ Left unchecked, these inaccuracies can pose a huge risk, especially in public services where trust and accuracy are critical.

Daniel Dredge also raised concerns around AI and its ability to potentially leave out critical information, asking ‘If you ask it to summarise an 80-page document and it comes out to one page, can it really be right?’ He continued, ‘It’s almost too good to be true. So I actually find myself going back checking to make sure it hasn’t missed anything. It almost sounds counterproductive.’ To mitigate these risks, it’s essential that staff are made aware of these potential pitfalls and remain diligent when reviewing AI-generated outputs.

While AI can be seen as a quick fix tool, it’s important local authorities are putting in the work ahead of time. Empowering staff with the right training will not only improve adoption but ensure AI is used safely, effectively, and ethically across local government services.

Public understanding and expectations of AI

While many people often use AI-powered tools in everyday life – like navigation apps or voice assistants – they often don’t recognise these as AI. So, when governments begin to implement more widely recognised AI solutions, this often leaves people feeling nervous or apprehensive, especially when it’s something they don’t have experience with themselves.

Panel host Cliff Dean, Chief Information Officer at North Kesteven District Council and West Lindsey District Council raised the question of whether the public truly understands how AI is being used in their lives and services. This led to a discussion around public trust, clarity, and the expectations around AI in different situations.

Madeline Hoskin talked about the different kinds of services local governments provide – from routine, low-risk tasks like bin collection updates to high-stakes support in health and social care. She explained that while the public may accept AI when it comes to transactional matters, they would expect and prefer human involvement in emotionally significant or vulnerable situations. She noted, ‘Would I want an AI making a decision about my grandmother’s care? Probably not. I want humans. I want hands, eyes and hearts involved in care.’

The panel agreed that AI is best introduced first in low-risk, back-office areas – automating repetitive tasks to improve internal efficiency – before being more widely used in public-facing services.

Daniel Dredge also added that public acceptance is largely based on positive experiences. If an AI-driven interaction is helpful, protects user data, and meets their needs, most people are likely to be comfortable with it. He explained, ‘Most people now are probably expecting to encounter it when they’re using digital services… But there’s quite a lot to do to get us in a position where we can go full-on customer-facing.’

Join our future discussions

Thank you to our panellists for their expertise and thought-provoking insights. If you’d like to take part in future discussions register your interest to join us at LGPN on 22-23 May!