Designing AI for better futures: From citizen friction to empowerment
AI can fix the drag felt by citizens when accessing multiple government agencies, but only if we design it with ongoing human accountability.
Sam Daley

Artificial intelligence offers new possibilities to ongoing efforts to redesign how public services create value. It can accelerate the shift from siloed, passive interactions to proactive, citizen-centred systems that guide users, anticipate needs and build trust. Proactive, personalised government services aren’t a new idea, but emerging AI capabilities now make this vision more achievable, where past efforts have failed to fully deliver on their promise.

The recent Mandarin/Liquid “Let’s get real about AI” survey of public service attitudes towards AI showed the public sector’s priority for artificial intelligence is justifiably focused on operational efficiency: faster processing, less administrative burden, and more time spent on the tasks that matter. Those outcomes are foundational, but if we stop there, we miss the real opportunity to reimagine how public services create value for people.

Public servants themselves prioritise reducing administrative burden (42% said this was the single most important place to invest in AI), but they also see the potential for AI to improve fairness, responsiveness and navigation for citizens. That means focusing on the public good as the goal of improved productivity.

Persistent frustrations show where to start
The top customer-facing challenges identified in The Mandarin/Liquid “Let’s get real about AI” survey results underline something every citizen knows: the biggest frustrations in dealing with government long predate AI. Long waits. Confusing eligibility rules. The Groundhog Day of repeating the same details to different agencies. These are structural pain points that technology has so far only chipped away at.

But they are also a roadmap. Each friction point is an opening for a different kind of service — one that is proactive, seamless and fair. AI gives us a powerful new tool in our arsenal as we move from reactive problem-solving to systemic redesign.

Fixing the invisible machinery lays the foundations
Behind every public interaction is a web of administrative work that is often unseen but acutely felt by the customer. Data is collected, checked, and handed off between systems and agencies. Citizens feel the drag in wait times and repeated questions. Public servants feel it in redundant processes and fragmented records.

AI can streamline this machinery. Natural-language models can automate routine verification and classification tasks. Machine learning can predict eligibility and flag anomalies before a citizen ever lodges a claim. Done well, these changes are invisible to the user but transformative in impact. The goal is not simply to “speed things up” but to erase the need for citizens to carry the burden of bureaucracy from counter to counter. At scale, the opportunities for impact include expanded reach through more fine-grained and nuanced digital services, capable of understanding and servicing people who might otherwise fall between the cracks.

Designing the glue, not just the pieces
The value of improving back-end processes is multiplied when we view it as a foundational enabler for the frictionless experience people expect. Public servants recognise this, with 82% of survey respondents saying they believe AI could provide intelligent guidance to help users find the right service. People judge services by the experience between interactions as much as by the interactions themselves. They are just as frustrated by what a service doesn’t do as they are with the things it does poorly.

This is where people’s interactions with AI are already changing expectations. Think of the leap from a keyword search to a ChatGPT-style conversation. It doesn’t hand you off to a page of content and leave it for you to find out what’s relevant to your question. It distils and summarises the information for you in a way that feels like someone making a genuine attempt to simplify things. It gives people the feeling of being understood. It’s also the ability to draw upon a breadth of knowledge without hopping across dozens of sites, skimming, backtracking, and trying again.

Those shifts in user expectations are profound and at odds with how government agencies traditionally offer information and digital transactions. Citizens will increasingly wonder why government services can’t anticipate intent, guide them step by step and offer context beyond a single agency’s remit.

For public services, the lesson is clear: interaction means more than clicks, forms and traditional information architecture alone. It means an active, two-way connection that educates, supports and empowers.

Trust is the differentiator
The breadth of knowledge underpinning language models like ChatGPT is powerful, but it comes with a trust problem. Generative models are notorious for delivering hallucinated errors with convincing confidence. In the commercial internet, users shoulder the burden of discerning what’s reliable. Government, health, and education agencies cannot offload that responsibility, but instead have an opportunity to set the benchmark through their expertise in designing for trust.

Here lies the public sector’s unique advantage. Public institutions hold a wealth of rigorously validated content. That “traditional information architecture” becomes a core strength. If AI can help navigate, combine and personalise that trusted material, citizens get the best of both worlds: the immediacy and adaptability of a smart personalised system capable of complex reasoning, anchored by authoritative sources and the guardrails of responsible system design.

Designers of digital services must treat exploring this advantage as a core responsibility. Simply making information available and accessible is no longer enough. We need a bolder vision for the quality of service experience needed to make a genuine impact in people’s lives, and must responsibly harness the power of emerging technologies when they offer breakthroughs in the scale or depth of that impact.

From silos to holistic service
All of this points to a bigger design challenge. Citizens do not live their lives according to agency boundaries, yet government services are still organised around them. People expect guidance that recognises their personal context and the broader ecosystem of support.

Imagine a mental health service that doesn’t stop at what it provides directly, but has trusted pathways for proactively connecting a person to housing support, education opportunities or community programs. AI can make that kind of cross-agency orchestration feasible, but only if we design for it – building systems that share data responsibly, respect privacy, and measure success by outcomes in people’s lives, not by the performance of a single department.

A call to designers and policymakers
Designing AI for the public sector is not (just) a technical exercise. It is a civic one. The choices we make now will shape how citizens experience government for decades. We must resist the temptation to limit our concept of AI to a bolt-on efficiency tool. Instead, we should view its potential as a catalyst for reimagining what holistic government service delivery looks like.

That requires multidisciplinary teams: designers who make sure solutions are focused on real human needs, technologists who can harness emerging tools, and policymakers who drive progress through an active approach to governance as an enabler. It requires investment not only in data infrastructure but also in human-centred design, testing and ongoing accountability.

The promise is worth it. AI can help us move from friction to empowerment, from isolated transactions to lifelong journeys of support. If we ground our work in public value and trust, we can create digital services that feel as seamless, connected and distinctly human as the rest of our lives.

Designing AI for better futures: From citizen friction to empowerment

Sign up to our newsletter.