AI Is Changing Government. But Process Still Comes First.

AI Is Changing Government. But Process Still Comes First.

There has been no shortage of discussion about AI over the past year. From copilots and chatbots to autonomous agents, the conversation has largely centred on what the technology can do.

But after spending time at Pega-Cognizant last week with technologists, public sector leaders and practitioners, one message stood out above all others.

The biggest challenge facing government isn't AI.

It's how government works.

The discussions covered everything from agentic AI to automation, but they repeatedly returned to a familiar theme: technology is rarely the hardest part of transformation. The real challenges lie in processes, governance, organisational culture and accountability.

One demonstration illustrated this well.

Today, redesigning a government service can take 18 to 36 months before organisations even agree on what should be built. Policy teams, operational leaders, technology specialists and delivery partners all bring different priorities, often spending weeks or even months aligning around a shared understanding of the problem before development begins.

AI has the potential to change that.

By rapidly generating visual representations of services, workflows and dependencies, organisations can explore ideas, challenge assumptions and build consensus much earlier in the design process. Instead of waiting months for documentation and process maps, stakeholders can begin reviewing and refining them within days.

The important point, however, isn't that AI is designing government.

It isn't.

People still define the policy intent, business rules, governance and desired outcomes. AI simply helps organisations understand complex processes more quickly, making it easier to improve services before significant time and money are invested in implementation.

That's a much more useful conversation than simply asking, "Where can we use AI?"

Perhaps one of the most thought-provoking questions raised during the event was:

"If it's too good to be true, what's the catch?"

It's a question worth asking.

Every technological advance comes with trade-offs. Faster design still requires robust governance. Automation still requires oversight. New capabilities still demand investment in integration, security, skills and organisational change.

Technology can make transformation easier.

It cannot do the transformation itself.

Another observation that resonated was that AI is no longer something organisations are planning for, it's already becoming part of everyday work. The question is shifting from whether organisations should adopt AI to whether they are adopting it thoughtfully. That distinction matters.

Too often, organisations ask how AI can improve their services before asking whether those services are designed effectively in the first place. AI doesn't fix broken processes. It exposes them.

If a process is overly complex, poorly understood or burdened by years of accumulated workarounds, introducing AI won't solve those underlying issues. It may simply enable the organisation to execute an inefficient process more quickly.

One analogy shared during the discussions captured this perfectly.

If you buy a new electric corkscrew but your partner the intended user was not involved in the decision and has never used one before, will they use it if the old corkscrew is still in the drawer, or will they default to what they already know?

Digital transformation isn't just about introducing new technology. It's about understanding why existing processes evolved, redesigning them around today's needs and giving people the confidence to work differently. Otherwise, organisations risk layering new technology over old ways of working.

That challenge is particularly relevant in government. Public services cannot simply optimise for speed. They must also maintain trust, transparency and accountability.

The discussion around AI as a "co-worker" rather than a replacement for people reflected this balance well. Used appropriately, AI can reduce repetitive tasks, help navigate complexity and allow employees to focus on work that requires judgement, experience and human interaction. But it also raises one of the defining questions for public services over the next decade.

If AI increasingly supports decision-making, who remains accountable?

For government, accountability cannot simply become "the computer's decision." If an AI recommends an outcome that a civil servant cannot fully explain, who is responsible for approving it?

The phrase "human in the loop" has become increasingly common, but it also deserves greater scrutiny. Is it enough for someone to approve an AI-generated recommendation, or does meaningful accountability require understanding why that recommendation was made?

These questions are unlikely to slow AI adoption, but they should shape how it is implemented.

One government representative spoke about the importance of "security by design" embedding security into technology from the outset rather than treating it as an afterthought. The same principle applies more broadly. Governance, explainability and accountability need to be designed into AI-enabled services from the beginning, not added later.

That, ultimately, was the most important takeaway from the event.

AI is undoubtedly changing government, but technology alone will not deliver better public services.

Successful transformation will continue to depend on understanding processes, designing services around citizens and building organisations that are ready to change alongside the technology they adopt.

AI will continue to evolve at remarkable speed.

Good judgement, accountability and public trust will remain human responsibilities.

And those are the foundations on which successful government transformation will continue to depend.