Governments struggle to adopt AI because legacy systems block the way
Public sector leaders across multiple countries face a common problem: they want to use artificial intelligence to improve government operations and citizen services, but outdated IT infrastructure, siloed data, and undertrained staff are stopping them.
At a roundtable discussion held at Global Government Forum's Innovation 2026 conference, government officials and experts identified three categories of barriers-budgetary, political, and organisational-that prevent modernisation projects from getting off the ground.
The legacy IT problem
One UK participant described a "legacy mindset" among IT teams that favours keeping old systems running over adopting new technology. Risk-averse attitudes, though well-intentioned, entrench outdated infrastructure rather than replace it.
The same official noted that while other countries have begun tackling legacy systems, the UK remains "still running essentially a legacy of pre-AI IT services".
A recurring issue: promising AI projects lose momentum during rollout. As one attendee put it, "Something dies on the way… when we're actually trying to get things rolled out".
The business case problem
Funding decisions often stress avoiding losses rather than achieving better outcomes. One participant described evaluating a border control modernisation programme that would cost hundreds of millions. The business case succeeded because officials framed it as: without modernisation, the border becomes insecure, so the return on investment is infinite-you cannot put a price on preventing a terrorist attack.
This logic secured funding for that project. But speakers agreed government needs a more practical framework for funding AI projects where the stakes are lower.
The gap between policy and delivery
UK attendees said political signals to innovate with AI are clear at the top. The problem is translating that direction into local action.
"The signals from the top are, do it - let's get the technology," one official said. "It's how that then translates into support locally for actually making things happen. And I think there is a little bit of a gap there."
Organisational resistance
Officials acknowledged that rethinking organisational structures to support AI adoption is necessary but still in early stages. One attendee warned against "organisational resistance to AI - and not thinking big enough about it."
The opportunity, they said, is to "really transform and rethink the way that we do business" and change how government interacts with citizens. "I think people haven't quite latched on to that."
Learn more about AI for Government and how organisations are addressing these adoption challenges.
Your membership also unlocks:
