Nine
in ten UK local councils are already using artificial intelligence or
actively exploring it, according to a survey conducted by the Local
Government Association between December 2024 and February 2025. That figure
would have seemed implausible three years ago. For an institution not
typically associated with rapid technology adoption, the scale of engagement
is remarkable, and the pace is accelerating. The experiments are not being
announced with fanfare or launched with dedicated press releases. They are
being run, often quietly, by teams managing planning applications, pothole
repairs, social care referrals, and waste collection routes. AI has arrived
in local government not as a grand transformation programme but as a
practical response to a very old problem: more demand for public services
than the available budget can comfortably meet.

The LGA’s
State of the Sector report on AI found that councils are deploying
tools across highways maintenance, social care, planning enforcement, and
citizen-facing services. The variety reflects local councils’ position as
generalist public organisations responsible for a wider range of services
than almost any other institution in British public life. When AI can be
applied to a process, local government has a process for it.

The Financial Case That Is Driving Adoption

The backdrop to the experimentation is a fiscal environment that
has left many councils managing rising demand with static or declining
real-terms budgets. The financial case for AI is correspondingly compelling.
Analysis by the Heriot-Watt University study of AI
readiness across 208 UK local authorities found that AI is being
used to handle tasks that previously required skilled officer time, from
automated sorting in recycling facilities to AI-enabled pothole detection
that allows road teams to identify defects without manual inspection of every
surface. Where the tool works reliably, the return on investment is
measurable in reduced staff time and faster service delivery.

Tony Blair Institute analysis cited by multiple council
procurement teams models AI as capable of improving or automating approximately
26 per cent of council tasks, saving around one million staff hours and
roughly 30 million pounds per year in a single authority. Scaled across
England and Wales, that modelling suggests up to eight billion pounds in
annual savings. These figures are projections rather than observed outcomes,
and the experience of technology adoption in public services counsels caution
about projections made before implementation. But they reflect why the
financial case is taken seriously, even by councils that remain cautious
about pace.

What the Experiments Look Like in Practice

The practical applications are more varied than headline numbers
suggest. In social care, Medway Council’s deployment of Lilli, an AI-driven
in-home monitoring system using sensors to identify changes in residents’
daily routines, has produced around 1.6 million pounds in care cost savings
by enabling earlier intervention and reducing avoidable care escalations. In
planning, councils including Hillingdon and Exeter have piloted AI tools that
can digitise historic planning records at speeds that reduce processing time
from one to two hours per file to approximately three minutes, compressing
backlogs that have accumulated over years of underfunded manual
processing.

Traffic management is another active area. Adaptive traffic signal
systems being tested in international comparator cities are producing
reductions in commuting times of more than 20 per cent in pilot conditions,
by processing real-time camera and sensor data to adjust signal timing
dynamically rather than running on fixed schedules. These examples sit
alongside more prosaic applications: AI-assisted processing of citizen
enquiries, natural language processing applied to public consultation
submissions to identify recurring concerns, and predictive maintenance
scheduling for council-owned assets.

What This Means for You

For residents, the most direct effects of council AI deployment
are visible in service responsiveness. When AI tools process planning
applications faster, decisions that previously took months can be returned in
weeks. When predictive maintenance identifies a road defect before it becomes
a pothole, the intervention is cheaper and less disruptive than repair after
failure. When social care AI enables earlier identification of residents
whose needs are increasing, the intervention can be less intensive and less
expensive than a crisis response.

The less visible effects are more complex. As our coverage of
AI
in democratic governance found, the question of accountability when
automated systems inform or make decisions about public services is not
resolved by the efficiency gains those systems produce. A planning decision
informed by AI analysis of consultation responses is still a decision that a
human officer is responsible for, but the transparency of how AI shaped the
officer’s judgment matters for the legitimacy of the outcome. Councils that
have moved fastest on AI deployment have generally also invested in
transparency frameworks that explain to residents how AI is being used and
what role human judgment plays in consequential decisions.

The Governance Gap

The Heriot-Watt study found substantial variation in AI readiness
across UK councils, shaped by differences in leadership priorities, legacy
technology infrastructure, and the pace at which national guidance is
developing. Many councils remain at very early stages of building the data
foundations that AI tools require to function reliably. The risk of the
current period, in which enthusiasm for AI’s potential is high and governance
frameworks are still being developed, is that councils adopt tools faster than
they can evaluate their effects or establish meaningful accountability for
outcomes.

As our analysis of how
organisations build ethical frameworks for AI deployment found, the
gap between stated commitment to responsible AI and operational governance of
how systems behave in practice is where risk accumulates. In local government
that gap has particular significance: the people most dependent on council
services are often those with the least power to contest decisions they
believe are wrong. The responsibility to close that gap sits with both
councils and the national bodies providing guidance, and the current pace of
deployment suggests it cannot wait for governance to be fully resolved before
experimentation continues.

The picture that emerges is of a public institution in genuine
transition, driven by financial pressure more than technological enthusiasm,
and finding that AI can do real things for residents and officers when it is
applied thoughtfully to well-defined problems. The regulatory
frameworks that will govern AI in public services are still being
shaped. What is already clear is that the question for local councils is no
longer whether to engage with AI but how to do so in ways that are effective,
accountable, and fair to the communities they serve.

The variation across councils also reflects differences in the
scale of the data challenge that precedes AI deployment. AI tools require
clean, structured, accessible data to function reliably; they inherit the
quality of the underlying data infrastructure and amplify its limitations as
well as its strengths. Councils that have invested in modernising their data
management over the past decade are finding AI adoption substantially easier
than those still managing records in legacy formats or across fragmented
departmental systems. This means that the councils with the most to gain from
AI, those operating under the most severe budget pressure with the least
mature digital infrastructure, are often the ones least able to adopt it
quickly. Addressing that gap requires investment in data foundations
alongside investment in AI tools, and the two are frequently separated in
funding programmes that treat AI adoption as a distinct initiative rather
than as the downstream benefit of better data management.

About the Author

Stuart Kerr is the Technology Correspondent for LiveAIWire. He
writes about artificial intelligence, emerging technology, and the forces
reshaping work, business, and society.