Nearly Right

The efficiency delusion: why AI cannot fix Britain's broken councils

How the £8 billion productivity promise exposes the dangerous gap between technological fantasy and democratic reality

When Birmingham City Council effectively declared bankruptcy last September, it joined a growing list of local authorities overwhelmed by impossible mathematics. Europe's largest council could no longer balance the books for more than a million residents. Yet within months, the Tony Blair Institute was promoting a different kind of calculation: artificial intelligence could deliver £8 billion in annual productivity gains across English and Welsh councils, equivalent to £325 per household.

The promise sounds compelling until examined through the diagnostic lens of what actually broke these institutions. This isn't a story about inefficient processes waiting to be automated. It's about a deliberate political choice to starve local government of resources whilst demanding ever-expanding services—then selling technological solutions to problems technology cannot solve.

Diagnosing the productivity mirage

The Tony Blair Institute's analysis rests on a fundamental misdiagnosis: that council struggles stem primarily from administrative inefficiency rather than systematic underfunding. Their researchers partnered with a single local authority to map staff tasks, found that AI could "automate or improve" 26% of activities, then extrapolated this across all 317 English councils to reach their £8 billion figure.

This represents textbook technological solutionism—the belief that complex systemic failures can be reduced to engineering problems. The institute's methodology reveals the flaw: they measured task automation, not outcome improvement. A social worker spending less time on paperwork delivers value, but it doesn't address why their caseload doubled during austerity, why their department lost a third of its budget, or why the families they serve face longer waits for diminished services.

Real implementations expose more modest capabilities. Kingston Council's Magic Notes system, developed by social enterprise Beam, reduced administrative time for social workers by 63%. This helps overworked staff, but operates within existing resource constraints rather than addressing the structural problems that created those constraints. The technology optimises existing capacity—it cannot restore the capacity that austerity removed.

The arithmetic of systematic failure

To understand why AI cannot solve councils' crisis, examine the actual system breakdown. Between 2010 and 2020, councils lost more than 50% of their government grants in real terms. Core spending power remains 10% below 2010 levels even after recent increases. This isn't marginal inefficiency requiring productivity improvements—it's fundamental system redesign through resource withdrawal.

The operational consequences are visible across Britain's high streets. Nearly 800 libraries closed between 2010 and 2019. Youth services budgets fell by 70%. Sure Start centres disappeared from neighbourhoods where children most needed early intervention. Meanwhile, demand for statutory services—adult social care, children's protection, homelessness support—continued rising with an ageing population and growing inequality.

Birmingham's trajectory illustrates impossible system dynamics. The city lost £736 million in government funding whilst its population grew by 7.5%. Even perfect administrative efficiency couldn't bridge that gap. The council's bankruptcy wasn't caused by inefficient note-taking or slow procurement processes. It resulted from a decade of being asked to deliver expanding services with systematically reduced resources, until the system mathematics collapsed.

The consultant dependency syndrome

The same period that saw councils starved of resources witnessed an explosion in spending on external consultants and digital transformation projects. The public sector spent £14.5 billion on contractors, managed services providers and IT consultants in 2023—55% of total digital spending, whilst permanent staff received just 20%.

This reveals the central systems failure in the current efficiency narrative. Councils supposedly lack internal capability to deliver services effectively, justifying both austerity cuts and expensive external interventions. But the resulting dependency on consultants costs far more than maintaining internal capacity would have done, whilst leaving councils without the institutional knowledge and democratic accountability that permanent staff provide.

The pattern repeats with each technological promise wave. The National Programme for IT consumed £10 billion before abandonment. The courts digitisation programme ran three years late and over budget. Brexit preparations saw £1.3 billion flow to consultancy firms in seven months alone. Each failure triggers the same misdiagnosis: government needs disruption, new technology, business-like efficiency. The structural causes—political short-termism, procurement dysfunction, democratic deficit in technical decision-making—remain unaddressed.

Democratic governance versus algorithmic administration

The deeper systems failure lies in what the AI efficiency narrative obscures about democratic governance. Local councils don't exist primarily to process tasks efficiently—they exist to make collective decisions about shared resources and competing priorities. The seemingly inefficient processes of democratic deliberation, public consultation, and political debate aren't system bugs to be automated away. They're the core features that distinguish legitimate government from technocratic administration.

When Kingston Council's social workers spend less time on paperwork, that creates value. But when that efficiency gain justifies further budget cuts rather than improved service quality, the technology becomes complicit in the broader system hollowing-out. The £8 billion in supposed savings could easily become £8 billion in further cuts, leaving councils with the same impossible pressures but fewer staff to manage them.

International experience provides diagnostic warnings about this trajectory. Algorithmic decision-making in welfare systems has consistently demonstrated biases against already marginalised groups. Predictive policing reinforces existing discrimination patterns. Automated planning decisions favour revenue-generating developments over community needs. Without robust democratic oversight and genuine resource adequacy, AI tools risk entrenching the inequalities that public services exist to address.

System restoration requires accurate diagnosis

Britain's councils are failing not because they're insufficiently automated, but because they've been systematically defunded whilst expected to solve problems created by national policy choices. Homelessness rises when housing becomes unaffordable, not when housing departments process applications slowly. Children enter care when families lack support, not when social workers write notes inefficiently. Adults need care as they age, regardless of assessment completion speed.

The £8 billion AI promise offers technologically sophisticated magical thinking—the belief that better tools can substitute for adequate resources and democratic accountability. It's particularly seductive because it implies improvement is possible without confronting the political choices that created the current system failure.

Genuine system restoration would require acknowledging that local government needs sufficient funding to deliver services communities require, democratic structures that prioritise community needs over contractor profits, and technological tools that enhance rather than replace human judgement and care. The efficiency gains AI can deliver are valuable marginal improvements to a system requiring fundamental repair.

The danger isn't that AI will fail to deliver promised savings—though it likely will. The danger lies in technological solutions distracting from addressing structural problems no algorithm can solve. Birmingham didn't need better note-taking software. It needed sustainable funding for essential services and democratic control over its own budget.

Until Britain's political leaders find courage to diagnose the real causes of local government's crisis, no amount of artificial intelligence will restore the capacity that deliberate policy choices removed. The arithmetic of democracy, unlike the arithmetic of automation, cannot be optimised away.

The future of British local government depends not on smarter algorithms, but on whether we still believe communities deserve adequately funded public services controlled by the people they serve.

#artificial intelligence #politics