UK government plans for AI must follow established principles

User research, solid metrics and a focus on getting maximum value from existing data will all play critical roles in the AI-enabling of public services, writes Massimo Merlo at Elastic

 

In the first half of 2025, the UK government under Sir Keir Starmer made clear its commitment to harnessing AI for national renewal. Starting with the January launch of the AI Opportunities Action Plan, followed by the AI Playbook in February, and a ‘test and learn’ funding approach in March, these initiatives aimed to accelerate AI adoption across public services while ensuring trust and oversight.

 

Technology Secretary Peter Kyle emphasised the shift away from outdated processes that stifle innovation, stating that new changes would make innovation the default in government. With major investments – including a £1 billion boost to AI computing capacity and plans for an exascale supercomputer – the UK can drive AI development at scale to transform public services for citizens.

 

 

Trust is a vital consideration

This is bold, uncompromising talk, and the commitment expressed by government ministers is certainly encouraging. But there is one vital consideration that must accompany and underpin every effort made to deliver on their pledges: public trust. 

 

In short, for AI-powered public services to be accepted, the public must be confident that the technology is safe, secure and well-managed. And for such services to be adopted with enthusiasm, they must be well-aligned with social priorities and deliver a user experience that represents a significant step forward from previous government interactions that a user might have had. 

 

Whether they are registering a birth, applying for state benefits or renewing a passport, members of the public want to experience processes that are more streamlined, less frustrating and which keep them updated on the progress of their request. These services should be designed alongside the people that use them, and not just on their behalf, so that they truly respond to public needs. In an ideal world, they would offer the same simplicity and convenience that an increasingly digitally-savvy audience experiences when using private-sector, consumer-focused services to hail a ride or order a takeaway. 

 

Where possible, members should be able to ask simple questions without prolonged waiting times and decision backlogs, regarding their entitlement to social housing, for example, or whether they need to register their small business for VAT. The deployment of AI-enabled chatbots, agents, and self-service portals could dramatically turn around the current situation of delays and bottlenecks, enabling them to get rapid answers and complete transactions at their own convenience. 

 

 

A matter of metrics

All this is a tall order, of course – but it’s where the UK government needs to aim if it wants to win public trust and satisfaction in an era when user expectations of digital services have never been higher. 

 

To measure its success, what is needed is a standardised, joined-up approach to quantifying the benefits of AI projects, both in terms of better outcomes for users of these services and improved governmental efficiency. After all, civil servants also want more convenient and less complicated ways to get work done.

 

The AI Playbook suggests that AI project benefits might be measured in various ways: faster delivery of services, reduced staff workload, improved quality of services, and reduced reliance on humans to make decisions or perform specialised tasks. It also urges government workers responsible for building AI services to engage with people both inside and outside of government, to understand “their needs, values and priorities” and “the way the humans who will use the products or services behave, think and feel.” 

 

The development of metrics that measure not only how well the technology performs but also how well it meets internal targets and users’ needs, according to the Playbook’s authors, are urgently needed and have the capacity to support the ongoing improvement of AI services. 

 

 

Ageing technology

It’s also reassuring to see the government acknowledge that ageing technology potentially represents an impediment to its plans, as could data that is locked away in legacy IT systems. Its own research finds that one in four of the digital systems currently used by government bodies is outdated and costs the taxpayer some £45 billion annually in lost productivity.

 

This is an important point, because effective AI must be fed by data that is clean, up-to-date and held in accessible formats and systems. However, as many technology experts will testify, this typically does not constitute a reasonable or economically valid argument to rip out and replace such systems wholesale. In many situations, there are better ways to approach the issue, which help preserve existing investments. 

 

More specifically, what is needed is a way to aggregate data held in a wide range of disparate systems so that it can be searched and analysed as a unified ‘whole’. A centralised approach to bringing together data from old and new systems can be achieved using open-source technologies, making it more cost-efficient. And, as new targets for AI enablement emerge, this approach allows for new analytical workloads to be developed and deployed incrementally, rather than developers having to start all over again from scratch. 

 

Mr Kyle’s call to “make innovation the default” is timely and inspiring, but success will depend on grounding that innovation in proven principles. By prioritising user needs, applying rigorous metrics, and harnessing the full potential of existing data through a centralised approach, the UK can realise its bold AI vision responsibly and effectively, delivering public services that truly work for everyone.

 

---

 

Massimo Merlo is AVP of UKI at Elastic

 

Main image courtesy of iStockPhoto.com and Gwengoat