So far, the focus on AI in government has been around cost saving, producing better decisions and doing more for less. We’re already seeing AI tools being used in processes and day-to-day tasks. Ameet Gonsai and Gwyn Calley, designers at Transform, recently spoke about how AI can assist a more efficient design process, for example, and I’m an avid user of AI assistance on Miro. But how does the citizen and the state factor into this? The conversations around efficiency now aren’t unlike those of the early 2000s exploring the potential of the internet to do things differently, with a focus on productivity and marginal gains. But what about the radically different ways of delivering services that will emerge due to the disruption brought by AI – like internet-enabled business models of the 2010s, such as Netflix?
Attention should turn to how government can use AI to disrupt delivery models and design better interactions between citizen and the state rather than wait and be catching up with the expectations of the citizens in the AI era. And to do that, government teams need to experiment, despite the many headwinds.
“Most of government is services most of the time.” When we design good service outcomes, we think about the interaction between agents and users, whether that’s by providing them information to deliver interactions, designing tone of voice, or even ensuring the right hold music is played as you wait in HMRC’s queue. Government will need to begin thinking about how to design ‘AI agent’ components in the same way, so that the way an interaction with an AI ‘feels’ is as good, if not better, than a human one.
The opportunity here isn’t just to replicate what already exists, but to explore how AI can create a step-change in that interaction, for example, by combining roles that traditionally sat separately, we may see AI agents interacting with each other to achieve a joined-up outcome for users.
As a citizen, can you imagine ‘speaking’ to an agent who can register your new child’s birth and provide you information on your maternity and tax benefits? Can you imagine how a border interaction could be better supported by technology that can bring together interactions from multiple departments. Patrick O’Shea’s recent post on transforming the border experience illustrates just how complicated this looks when you map how many different departments are present and the opportunity for AI to disrupt how this might feel to an end user.
All this points to the opportunity to proactively assist users through their journeys. We know that high-complexity users, with multiple government service interactions, under significant stress, struggle to interact with government, even digital by default services. There’s an opportunity for AI to proactively design and manage support around a user journey. For example, by noticing an appointment has been missed, designing a prompt, rebooking a service and offering support to ensure a user feels able to take up a service offer.
Identifying high-stress points and areas of tension in a journey isn’t new. At Transform, we’ve used AI and machine learning on large data sets to understand when and which junior doctors are likely to drop out of their training journey to find a different career, enabling Health Education England to target specific interventions to keep them. What is new, is the opportunity to simplify government for users, by tasking AI agents to join things up.
For government to be able to achieve this disruption quickly and for the benefit of users, it requires teams to start experimenting. There are lots of reasons not to do it – imperfect data, unknown benefits and outcomes and a lack of well understood frameworks, but government has the opportunity and incentive to help define those through doing. Perhaps the most important thing to develop to empower teams is through the language they all speak – the ‘AI aware’ service manual.
The service standard helped teams to adopt a consistent and high-quality way of delivering in the internet-era that put users at its heart. The sentiment should remain the same, do the hard work to make things easy and solve real user problems, keeping in mind AI could fundamentally change the shape of teams, the way services are assessed and good practice around service maintenance and performance.
In the future, teams across government will be contributing to and drawing from a ‘good AI design guide’ like the current Government Design System. Content design, for example, is likely to change and the role of a content designer may well be about ‘teaching AI’ to deliver content to different audiences, in different ways. We know that business leaders, school admins, parents and vulnerable people seeking help all need to receive content in different ways, so how we design AI interactions that provide for that, will be a new opportunity area.
The challenges remain the same as developing any service for users in the public sector. Fundamentally, people remain at the centre. Can you imagine the complexity of designing an AI agent that meets the business needs of multiple departments? The ability to bring people together, align them and deliver outcomes that satisfy all will remain a key challenge in government, which is likely to keep designers and product people busy.
Data, of course, is key to building successful AI tools, but that doesn’t mean large data transformation programmes are strictly necessary for teams to experiment with and develop new capabilities. In fact, leadership ought to be asking ‘what can we do with what we’ve got?’
Measuring impact. It’s a hobby horse of mine, but if we can’t use AI to improve the way we measure impact of the services we design and build, we’ll struggle to justify exploring and investing in new technology when budgets are tightened. Sure, getting a claim processed faster is necessary and impressive, but how can we evidence AI contributing to solving the knotty problems at the heart of these services? Issues such as providing an excellent teacher for every child, ensuring every child can achieve their potential, providing greater access to fair and equal justice outcomes or protecting the environment and achieving net zero.
At Transform, we’re exploring how AI can enable and disrupt new ways to deliver services and businesses. Get in touch at transformation@TransformUK.com if you want to find out more and we’ll put you in contact with one of our experts.
