Ensono: Government must be clear on AI goals or risk expensive failure

The UK government’s ambition to lead in public sector AI adoption could falter unless it starts with a clear sense of purpose. That was the warning from tech industry veteran, Phil Karecki

In a discussion on climate-focused AI, the current insurance technology leader at IT managed service provider (MSP) Ensono argued that while AI offers “massive potential productivity savings,” the reality is far more complex.

“Understand the outcome that you want. I don’t care what you do, first understand what your end goal happens to be,” he urged public sector organisations. “This is why digital projects fail 70 percent of the time – they didn’t know what it was they wanted to achieve.”

Karecki said governments too often specify the system they think they need, only to find it fails to deliver on real-world priorities. “I had a customer once tell me, ‘you gave me exactly what I asked for. But not what I needed.’ And that’s the risk the government faces – that they’re going to build exactly what they wanted, but it’s not what they need.”

Karecki’s message to policymakers is to not collect data for its own sake, but to plan for its use from day one. “Someone dumps a bunch of stuff in your lap – great. But if I haven’t already planned for the consumption, if I haven’t laid the groundwork for those interconnections between private and public, between climate and building construction data… I’m sitting there holding something very, very valuable that I can’t do anything with.”

And when it comes to climate and AI, prevention is better than cure: “The interest has shifted from making sure that we’re all good when something happens to making sure something doesn’t happen. That’s the government’s responsibility with climate data.”

Karecki also stressed that AI success in the public sector cannot be separated from private sector readiness.

“We’re no longer at a point in our world where those two are as distinct as they used to be. They can’t be. We’re too dependent upon each other… what we’re doing in insurance, what we’re doing in banking… it’s no longer just a federal conversation.”

He likened it to car manufacturers’ safety pact: “When you come up with a new safety feature, you get use of it for a year or two, and then you share it with the rest of the industry… we all win.”

AI and climate: beyond the weather

On the climate front, Karecki stressed that AI’s role isn’t just about forecasting storms. “Climate at its core will give us the predictability… but it’s not simply climate. It’s understanding the impact of that data and what’s the value you as the government want to derive from it.”

For example, the ability to predict extreme weather is likely is useful – but AI’s real value comes from combining that climate data with construction records, deforestation patterns, or infrastructure weaknesses to anticipate knock-on effects like flooding or landslides.

Karecki also described using combined satellite, weather, and GPS data to warn boat owners of incoming storms – and even to alert rescue services when a vessel’s signal disappeared mid-emergency.

“Now you’ve taken an active role… you’re saving a life because you knew where they were, where the storm was, and where they were last seen,” he said.

On the physical backbone of AI – the datacentres themselves – Karecki also noted that governments are accelerating rollout plans but warned of environmental trade-offs.

“When you’re telling me I need a nuclear generator to provide the power for my datacentre, my gut reaction is I’m no longer meeting those [ESG] commitments that I’ve made,” he said.