Blog

When AI Becomes the Front Door to Government: Rethinking Digital Services for Changing Citizen Behaviour

Government

Data & AI

UCD

View more blogs

In this blog, Emily Allinson, Head of Capability Development at Opencast, explores the impact of new technology on citizens, the changing nature of their experiences, and the implications for how they engage with government. It takes a forward-looking view, based on the assumption that current trends continue and citizens increasingly turn to AI first when accessing government services and information.

The blog suggests that we may be at a pivot point in citizen behaviour and expectations, and that the user experience of government services could change significantly. Given that possibility, it's worth reviewing digital transformation roadmaps to ensure government is responding to changing user needs and emerging technology requirements.

UCD is key to understanding evolving patterns of citizen behaviour, emerging user needs and barriers to access. We'll need to define new service patterns, guidance and standards to help citizens access trustworthy, accurate and timely information and services.

Context

Changes in technology change how people access and engage with government.

A good example is smartphones, and the ‘mobile first’ era of digital transformation. Smartphones changed the way citizens could access the internet. Digital teams working in government then had to develop new design patterns and style guides to ensure all users got the online service they expect on mobiles. Government websites changed to accommodate those changes, resulting in new design patterns, content style recommendations, and accessibility improvements.

Coupled with social media, smartphones ushered in an era of citizen expectations of government being ‘always on’, with fast responses and resolutions, personalised services and digital touchpoints being the first port of call (rather than paper forms, telephone calls, in-person meetings etc.).

Around the same time, Government Digital Service (GDS) was set up to enable digital transformation in UK government and meet these changes in citizen expectations for a digital government.

We’re now at another potential pivot point. Insight from Gartner indicates that AI interfaces like ChatGPT could soon be the main way people access, engage with, and understand government guidance and services.

Even beyond AI chat products, most browsers show an AI summary at the top, meaning that many people engage with AI without actively seeking it out. These products and search summaries present information confidently, without people visiting a website.

What happens then, when someone relies on AI to find out whether they're entitled to a benefit, what evidence they need and how to apply, without ever visiting GOV.UK?

What user experience implications will this have, and how should government adapt its transformation roadmaps to better serve citizens who are making use of AI?

What UCD considerations should be made to account for changes in user experience, expectations, and behaviours?

What new standards, patterns, advice and training needs to be in place to address novel user needs, threats and contextual risks when citizens are “doing government” through AI?

Insights

We should be exploring how government services need to adapt to meet the needs of citizens as they are adapting their behaviours to use AI.

Current Government Service Standard documentation, especially assets like the design library and style guide, are specialised for web. But what happens if people don’t visit gov.uk-based sites, instead interacting with government purely through their AI solution of choice?

Patterns, standards, key messages, archiving policies, accessibility guidance, content maintenance plans, etc. will have to be adapted or evolve to support users accessing accurate and timely government content via AI.

UCD roles are central to understanding this change and its implications for citizen behaviour, as well as designing new ways to support citizens to access government services.

This should be a content-led evolution, and we’d expect to see content roles evolving to be focused on optimising government services for strategic AI interface consumption.

At the moment, there are several key ways that citizens use AI products to access government services:

Be better informed - Find information to inform decision making by using AI to summarise options, give advice on next steps, etc.
Make a decision about what to do - For example, to decide what to apply for first from a range of options.
Prepare - To collate, craft and refine applications, or get support drafting evidence statements, etc.
Troubleshoot a problem – To understand how to fix a problem, such as if a form field isn’t accepting an answer.

The first and second points are open to anyone using an AI interface and sit alongside other task-based AI interface usage like comparing holiday options or looking for recipe advice.

The third requires inputting personal data, and may include giving the models access to evidence like payslips, health records, previous correspondence with government departments, etc. This has obvious security, privacy and legal implications for citizens.

What we can’t do yet, is use AI to directly apply for government services on our behalf. So, there is still a human effort required to access government services, submit material and fill in forms correctly. But even the changes in finding information and collating application content could result in some significant shifts to the way citizens experience government services.

What This Change Might Mean

Here are a few thoughts on what the change might mean for government digital transformation plans. Note that these are not exhaustive, and some are speculative.

Big shifts in technology usage mean changes to user needs, expectations and behaviours.

We'll need to create new design and service patterns that support findability and prioritised presentation in AI interfaces.

Trust has always been a central concern for government services, and this is likely to change what is considered trustworthy information.

There will be an even greater emphasis on content strategy, maintenance and deprecation of old and inaccurate content. This will help ensure citizens using AI are more likely to be presented with accurate information, although there is a limitation to how much content designers are able to influence what is actually surfaced in an AI UI.

Plain English and clarity will be more important than ever, to support AI summaries reporting unambiguous recommendations and next steps.

This feels like an evolution of best practice for content and SEO.

The digital divide is likely to increase, and who it impacts will change. Some may be priced out of using AI, and there's a growing cohort of people who actively avoid using it due to concerns about sustainability and ethics.

Assisted digital, offline and what we might consider "traditional online" (for example, access via web browser) remain core requirements for government services. As AI adoption grows, service teams will need to ensure that alternative routes remain visible, accessible and equally effective.

There may be a shift/increase in service volumes if AI surfaces more eligible options for citizens and makes drafting applications easier.

Service architecture, etc. will need to be robust enough to cope with increased application rates. Internal teams managing cases and applications will need to be ready to cope with additional volumes (as well as continuing to clear known backlogs that are causing delays to citizens accessing services).

Automation of certain processes is already in progress. For example, chatbots process routine enquiries to reduce workloads for call centres. But this needs to be balanced against the reputational risk of citizens perceiving a lack of human oversight or suspecting unfair or biased decisions.

If more applications, appeals or correspondence are written with AI, that could change the skills and experience required of the staff processing them.

Conclusion

UCD specialists should start investigating and paying attention to the changes to citizen experience of government services. However, this isn’t just a responsibility for UCD professionals, it also relies on decision makers in government revisiting the core principles and assumptions of their current digital transformation roadmaps.

There are practical steps that different roles across government can take now to understand these emerging behaviours and prepare for their potential impact:

User Researchers – Add AI tooling, experience and behaviours into research plans.
UCD Leaders – Create departmental opportunities to join up insights and overarching user needs of citizens using AI.
Cross-government UCD teams – Create cross-cutting working groups to use insights to review design patterns and make changes/additions to design system and style guides.
UCD Practitioners – Test new patterns and guidance, feed back into design system for iterations. Look out for and test adaptations to emerging trends. Don’t rush to find solutions to a change that in many ways is still taking shape.
Service Owners – Create cross-cutting working groups to use insights to review service and product roadmaps.
Strategic Government Action – Review holistic digital transformation roadmaps with insights into citizen AI usage, to revisit plans and amend to meet emerging citizen needs and expectations.

The future impact of AI on citizen interactions with government remains uncertain, but understanding emerging behaviours now will help ensure services continue to meet user needs as expectations evolve.

Person using an AI chatbot on a smartphone.