With Patrick Vallance’s appointment to chair the prime minister’s new AI Taskforce in the UK signalling a shift in government AI from experimentation to delivery, the momentum needs to come from stronger central leadership. But is that enough?
The real test will be whether leaders can earn the trust of frontline staff, understand how it supports their work and have the confidence to change established ways of operating.
Public sector organizations are not starting from a blank page. They are working around legacy systems, fragmented data, heavy caseloads and years of incremental reform. For many employees, a new platform, process or digital tool can feel less like progress and more like another system to learn. Unless leaders confront that reality early, AI adoption may create the appearance of transformation without delivering meaningful operational change or a better experience for service users.
Adoption Is the Missing Measure of AI Transformation
For too long, adoption has been treated as the final stage of a transformation program: the system is built, training is scheduled and staff are expected to adjust. Yet adoption is where the promise of transformation either becomes part of daily work or fails to take root.
That distinction matters because public sector employees are rarely resistant to change. More often, they are cautious because previous program promised simplification but created extra steps or systems that did not reflect operational reality. Caution is not reluctance; it is feedback from the people who understand where services could break down.
As AI becomes more central to government delivery, the challenge is not whether public bodies can procure new tools or launch pilots. It is whether they can establish the clear rules, governance and confidence needed for people to use those tools consistently and responsibly.
AI changes the consequence of unresolved organizational issues because it can act on information at scale. Poor data, unclear ownership or fragmented processes do not simply reduce performance. They can accelerate the wrong outcome. In public services, where decisions shape budgets, service access and resident experiences, that risk must not be treated as a technical inconvenience.
Frontline employees who enter and maintain data are not always shown how it later informs service decisions or funding cases. When that connection is invisible, improving data quality can feel like administration rather than shared ownership of better outcomes. Leaders need to make that link clearer.
Trust Is Built Through Culture and Change Management
The modern public sector workforce is often discussed in terms of digital, data and AI skills. These matter, but they are not enough. A resilient workforce also trusts the direction of change, understands its purpose and has the confidence to keep improving as services, expectations and technologies evolve.
That trust is built through everyday leadership. Managers need to model the behaviors they expect from teams, challenge poor processes rather than recreate them in new systems and make data, governance and accountability clear enough so people act with confidence. These questions determine whether AI becomes an operational reality or more work.
Change management must also be built in from the start, not added once deployment is already under way. Employees need practical guidance, clear use cases and ongoing support. They need to understand how AI fits into wider processes and decision-making, as well as where human oversight remains essential.
This is particularly important in environments where outcomes directly affect services and those who use them. Basic training alone is unlikely to be enough when organizations are asking staff to adapt established ways of working while also developing confidence in new tools, new data practices and new accountabilities.
When change fatigue appears, it is tempting to view it as a people problem. In reality, it is often a design flaw. If transformation feels like another task added to a full day, people will find the route that feels safest and fastest. That may mean spreadsheets, manual checks or informal processes that continue long after a new tool has gone live.
The Measure of Success Is Whether Work Actually Changes
Government should stop measuring AI success primarily through launch dates and delivery milestones. These measures tell leaders whether a project has landed. They do not show whether it has been absorbed. More meaningful measures include whether old processes have been retired, decisions are being made faster, staff confidence is improving and workload pressure is reducing.
Before scaling an AI program, leaders should be able to answer four practical questions: which existing process will stop, who is accountable for the data and outcome, where human judgement remains essential, and how employees and service users will know that the change has made things better. If those answers are unclear, the organization is not ready to scale.
A technology rollout can be completed on time and still leave the organization fundamentally unchanged. Real transformation will be visible when staff stop needing workarounds, managers trust the new process and technology feels like part of the role rather than something running alongside it.
Patrick Vallance’s appointment should therefore be welcomed as an opportunity to bring clearer leadership and urgency to public sector AI. But the defining question for government leaders should not only be, “What technology are we introducing?” It should also be, “What needs to change in the way people work, and how will that improve public services?”
If AI is to deliver meaningful outcomes across government, then culture, user adoption and change management must be treated as core parts of the program rather than supporting activity. Stronger governance can set direction, but frontline confidence will determine whether that direction becomes lasting public service transformation.
