On the second day DigiGov Expo at the Excel, London, a packed Digital Foundations Theatre hosted its final, highly anticipated session to bring the conference to a close. Titled "AI is the Tool and Your People are the Key: Managing the Risk of Public Sector Legacy Modernisation," the discussion was chaired by Catherine Nikolovski, Executive Director of GovTech Global and the Civic Software Foundation. She was joined by an expert panellist lineup Ben Pert, Principal Engineer at Made Tech; Claire Potter, Chief Data Scientist and AI Ethics and Assurance Lead at HM Revenue & Customs; Jenny Booker, Chief Data Architect at the Government Digital Service (GDS); Richard Baynes, Deputy Director for Digital Delivery at DEFRA ; and Tom Collins, Principal Software Engineer at DVLA. The conversation moved beyond speculative headlines to examine the tangible realities, cultural shifts, and operational risks facing public sector technology and legacy modernisation.
Cutting Through the Hype to Focus on People and Culture
Opening the discussion, the panel addressed the wave of media discourse surrounding existential risks and automated doomsday scenarios. Consensus emerged that while frontier models are advancing at an astonishing pace, getting bogged down in catastrophic hypotheticals distracts from the real work at hand. Instead, public sector leaders must concentrate on their circle of control, prioritising the cultural shifts, leadership skills, and workforce capabilities needed to implement technology responsibly. Rather than focusing purely on what the technology can do, leaders were urged to focus on the human problems requiring solutions.
The Foundation of Data Sovereignty and Governance
A recurring theme throughout the session was that artificial intelligence cannot be viewed in isolation from foundational data infrastructure. Panellists noted that public sector organisations must secure reliable, sovereign core datasets rather than relying on unvetted external sources.
From an assurance perspective, discussions emphasised that while ethics alone will not prevent systemic failures, robust governance will. Accountability must remain firmly anchored to named individuals and departments. Automated tools, chatbots, or generative systems can never be held responsible for public service delivery errors, making transparency and clear lines of ownership vital for maintaining public trust.
Navigating Code Generation and the "Vibe Coding" Reality
The session explored how advanced AI models have drastically lowered the barrier to software creation, allowing non-technical business colleagues to prototype and build solutions rapidly. While this democratisation fosters diversity of thought and empowers departments to solve local problems, it introduces significant risks regarding shadow IT and long-term technical debt.
Debate touched upon the colloquial term "vibe coding" and its place in government delivery. While rapid prototyping and hackathons are invaluable for accelerating research, development, and cross-departmental empathy, a sharp distinction must be drawn between internal productivity tools and mission-critical public services. High-risk public-facing systems still demand rigorous software engineering discipline, comprehensive testing, and expert human oversight.
De-skilling Risks and the Future Workforce
As automation takes on routine development and administrative tasks, panellists raised valid concerns around potential institutional de-skilling. When engineers rely heavily on AI assistance, maintaining the critical faculty to review, challenge, and judge code quality becomes an essential skill. Furthermore, legal frameworks such as GDPR place strict requirements on human-in-the-loop accountability for automated decisions, meaning the civil service must actively cultivate critical thinkers who understand the mechanics behind the technology.
Conclusion
Concluding the final session of DigiGov 2026, the panel left the audience with a forward-looking challenge. Technology will continue to accelerate legacy transformation at an unprecedented pace, compressing iteration cycles and testing traditional delivery frameworks like the GDS lifecycle. Ultimately, the success of public sector digitalisation will not be determined by the sophistication of the algorithm, but by the empowerment, accountability, and capability of the people deploying it.
Liuba Pignataro
