AI can help councils close the gap - but only if it improves care, not just efficiency

The BBC’s reporting on Northamptonshire councils turning to AI amid significant financial pressures reflects a reality facing local government nationwide: demand is rising, budgets are constrained and adult social care remains at the centre of the challenge. But AI should not be viewed as a shortcut to savings. Its real value lies in helping councils make better, faster and more human-centred decisions.

Local authorities are being asked to do more with less while protecting the services residents rely on most. In North and West Northamptonshire, this has meant exploring AI-enabled transcription, digital tools to support independent living, automated initial enquiries and WhatsApp-based access to information. At the same time, the councils face a combined funding gap of more than £37 million for 2027/28.

These initiatives are understandable. They are also part of a much wider question for the sector: how can councils use AI responsibly to relieve operational pressure while safeguarding dignity, choice and trust?

The objective must be better outcomes

The conversation around public-sector AI too often starts with technology or cost reduction. It should start with outcomes.

In adult social care, that means asking whether a solution helps people:

• Access advice and support more easily

• Tell their story once rather than repeatedly

• Remain independent for longer

• Receive more timely assessments and care planning

• Communicate in ways that work for them

• Feel listened to, understood and in control

  
  

For frontline practitioners, it means reducing administrative burden so more time can be spent on meaningful conversations, professional judgement and coordinated support.

AI transcription is one clear example. Used well, it could reduce the time professionals spend typing case notes after conversations and improve the consistency of records. That time could then be reinvested in direct work with residents and families. But it must never remove the practitioner’s responsibility to check, correct and contextualise the record.

A transcript is not an assessment. An AI-generated summary is not professional judgement. And a completed workflow is not necessarily a good outcome.

Automation needs appropriate boundaries

The concerns raised by Age UK are important. In adult social care, an inaccurate transcript, misunderstood request or inaccessible digital interaction is not simply a technical issue; it could affect someone’s care, safety or ability to get help.

Councils need to distinguish between tasks that are appropriate for AI assistance and decisions that require accountable human involvement.

Appropriate uses of AI assistance	Uses requiring human accountability
Transcribing meetings and care conversations, with review	Eligibility decisions
Drafting routine correspondence	Safeguarding judgements
Signposting residents to verified information	Assessing mental capacity
Routing straightforward enquiries	Determining care packages
Identifying incomplete information in forms	Decisions involving risk, deprivation of liberty or funding
Supporting staff to retrieve policy and process guidance	Interpreting a person’s circumstances and wishes

The principle is straightforward: use AI to support people and streamline low-risk processes, not to make opaque or unchallengeable decisions about people’s lives.

Inclusion is a service-design issue

Digital channels can make services easier to access for many residents. A well-designed WhatsApp service, for example, may offer a familiar and convenient route for some people to ask basic questions or find the right team.

However, convenience for one group cannot become exclusion for another.

Older residents, people with learning disabilities, people experiencing mental ill health, those without reliable internet access, and people whose first language is not English may need a different route into support. A council’s AI strategy must therefore include accessible alternatives: telephone, face-to-face contact, advocacy, translated information and clear escalation to a person.

The test should be whether technology expands choice, rather than silently narrowing it.

Trust is built through transparency

Public confidence will be essential if councils are to scale AI responsibly. Residents should be able to understand:

• When they are interacting with an automated system.

• What information is being collected and why.

• Whether a conversation is being transcribed or summarised.

• How personal data will be stored, secured and used.

• How to request human support.

• How to challenge an outcome or correct an inaccurate record.

This is especially significant in social care, where data is deeply personal and often sensitive. Councils should not rely on generic assurances. They need clear governance, documented impact assessments, robust procurement standards, staff training and ongoing monitoring for errors, bias and unequal impact.

Transparency should be designed into the service from the beginning, not added as a compliance exercise once the technology is live.

From pilots to sustainable change

Local government does not need more isolated pilots that create short-term interest but fail to transform the experience of residents or staff. It needs a disciplined approach to implementation.

For Datnexa, responsible AI adoption in adult social care rests on five foundations:

1. Start with a service problem. Identify the source of delay, duplication, unmet need or workforce pressure before selecting technology.

2. Co-design with residents and practitioners. People who draw on care and support, carers, social workers, occupational therapists and customer-service teams should shape the solution.

3. Build human review into every high-impact process. The more consequential the outcome, the stronger the requirement for accountable human oversight.

4. Measure value beyond savings. Track staff time, waiting times, repeat contacts, service accessibility, resident satisfaction, quality of records and equity of outcomes.

5. Create governance that can keep pace. Establish clear ownership, data protection controls, model assurance, incident processes and regular evaluation.

A moment for purposeful innovation

The financial pressures facing councils are real. The BBC has reported that 218 upper-tier councils are collectively expected to make £3.2 billion in cuts, underscoring why innovation is no longer optional.

But financial necessity should not lower the bar for public service design. It should raise it.

AI can help councils reduce avoidable administration, improve access to information, spot patterns earlier and free up valuable time for the human work that matters most. Yet it will only deliver lasting public value when it is introduced with clear purpose, strong safeguards and a commitment to inclusion.

The most successful councils will not be those that automate the most. They will be those that use AI to make public services more responsive, more equitable and more human.