Here is a stat that should unsettle anyone responsible for the UK’s economic future: at the current rate of progress, it will take until the year 2102 for all UK AI adopters to reach the most advanced stage of use. Not 2032. The end of the century.
This is not a technology problem. AI task complexity is doubling every seven months. Agentic AI – systems that autonomously plan, reason through and execute multi-step workflows – has arrived. The problem is organisational. And nowhere is that gap more consequential than in the UK public sector.
We’re using smartphones to make phone calls
Nearly two thirds of UK businesses have adopted AI – up from 52 per cent last year – ahead of the European average, where just over half of businesses now use AI. Impressive. But more than half of those adopters are stuck at the basics: summarising documents, answering queries, scheduling meetings. Only 24 per cent have reached the advanced stages – building custom models, combining AI systems, or deploying autonomous agents.
The economic difference is not marginal. Advanced users report efficiency gains of 68 per cent. Basic users? 40 per cent. Closing that gap is worth an estimated £35bn in productivity gains by 2030 – roughly Manchester’s entire annual economic output.
The future of enterprise automation: A business leader’s guide to agentic AI
By Amazon Web Services (AWS)The public sector’s quiet lead
There is a lazy assumption that government is permanently behind on technology. The evidence says otherwise.
New research surveying 500 public sector decision-makers reveals that where government bodies have adopted AI, they are going deeper than the private sector. Thirty-one per cent of public sector AI adopters have reached the most advanced stage of use, vs 24 per cent across all UK businesses.
The use cases explain why. Public sector problems are inherently complex: multi-step casework, cross-departmental coordination, fraud detection across vast datasets. These are precisely the applications in which advanced AI delivers the most value. Sixty-one per cent deploy AI for large-scale data analysis. Fifty-seven per cent for workflow automation and case management. Forty-three per cent for fraud detection.
The Department for Work and Pensions (DWP) receives 25,000 letters every day – previously taking five weeks to process. Using generative AI to identify vulnerable cases on arrival and escalate immediately, DWP has achieved a 91 per cent success rate since June 2024. That is not a pilot. That is AI delivering for citizens at scale.
Why isn’t it scaling faster?
The barriers are structural, not attitudinal. Public sector leaders are clear-eyed about AI’s potential. But they are operating within systems designed for a different era.
Only 19 per cent have a formal AI strategy. Fifty-eight per cent have no dedicated AI budget. Forty-seven per cent cite annual budgeting cycles as a direct barrier – particularly toxic for AI, which requires sustained investment over years before it can deliver returns. And procurement frameworks that demand detailed specifications up front are sensible for buying desks but disastrous for AI, with which the best approach is to start small, learn and iterate.
These are not excuses. They are solvable engineering problems – but they require deliberate action.
Why this matters beyond Whitehall
When government adopts AI visibly and effectively, it transforms the wider market. Seventy-eight per cent of organisations say they are more likely to adopt AI themselves if the public sector leads. More than a third of start-ups cite government demand as critical to their ability to scale.
This is the multiplier effect. Government adoption builds public confidence, creates market demand, stimulates job creation, and raises ambition across the entire economy. Conversely, when government moves slowly, it holds back not just public services but the ecosystem of companies trying to build AI for complex problems. The UK has over 3,600 AI start-ups – Europe’s highest density – yet many still look abroad for the funding, infrastructure and demand they need to scale – a signal of how much value a faster-moving domestic market, with the public sector as an early adopter, could unlock at home.
Three things that need to change
The UK government’s own analysis suggests it could make more than £45bn in unrealised efficiency savings with the full potential digitisation of public sector services. That prize will remain theoretical unless three things happen:
Strategy must precede experimentation – but strategy is not the same as specification. Seventy-six per cent of the most advanced adopters have a formal AI strategy. This isn’t about writing detailed requirements before anyone touches the technology – that’s precisely the procurement trap. It’s about setting a clear, cross-departmental framework with defined objectives and governance, and then giving teams room to start small, learn and iterate within it. The goal is disciplined experimentation, not scattered pilots hoping for the best.
The skills crisis needs urgent intervention. Half of UK organisations cite digital skills shortages as their biggest barrier. It now takes eight months to fill a digital role – up 45 per cent in a year. Sixty-seven per cent of workers want AI skills but a third doesn’t know where to start. The appetite is there; the pathways are not.
Procurement and funding must catch up. AI does not work like buying equipment. It requires iteration, flexibility and multi-year commitment. Annual budgets and rigid specifications kill AI projects before they can demonstrate value.
The window is open – but not indefinitely
The UK has genuine advantages: world-class research, deep talent, Europe’s densest AI start-up ecosystem, and adoption rates that lead the continent. The public sector is not the laggard of popular imagination – where it has adopted AI, it is going deeper than most of the private sector.
But competitive advantage in AI is not static. Twenty-nine per cent of public sector organisations say they feel ready for next-generation AI. The question is whether the system will let them act – or trap them in a permanent state of ambitious inaction.
The UK has a head start. The question is whether we have the institutional courage to use it.
Related
