A quiet digital shift in South Kesteven
South Kesteven District Council has put it in writing. The council's ICT Strategy 2025–28 formally commits to assessing and incorporating artificial intelligence, machine learning, and automation into its service delivery — and, notably, has embedded an AI SWOT analysis directly into its Strategic Risk Register. This is documented policy, not a vague aspiration about future technology.
The context for that commitment is measurable. SKDC's Customer Experience Strategy recorded 461,171 visits to the council website in 2018/19, and set an ambition for 24/7 multi-channel access — web, chat, phone, face-to-face — alongside a clear shift in the skills expected of customer service staff, moving from telephony expertise toward digital competence. The direction of travel was already established before AI became a mainstream policy conversation.
The council covers a predominantly rural stretch of Lincolnshire: Grantham, Bourne, Stamford, Market Deeping, and the villages beyond. For residents in that geography, the local authority is not an optional service — it handles housing, council tax, planning permissions, waste collection, and a range of other decisions that directly shape everyday life.
Which raises a question this article follows to its end: when public services in a place like South Kesteven are shaped or executed by automated systems, who explains how a decision was reached — and who is responsible when the algorithm gets it wrong?
Which decisions, exactly?
Consider what a district council actually decides. Housing applications, council tax reduction claims, planning applications, bin collection routes, benefits referrals — these are the practical outputs of local government for residents in Grantham, Bourne, or the villages beyond. Any of them could, in principle, be shaped by algorithmic tools: eligibility scoring, triage systems, route optimisation, risk flagging.
The distinction matters because the stakes vary enormously. An algorithm that optimises which streets the bin lorry covers on a Tuesday is a logistical tool. An algorithm that helps determine whether someone qualifies for council tax support, or where they sit on a housing waiting list, is something qualitatively different — it shapes access to resources that for some residents are essential rather than convenient.
No published document sets out which of these service areas in South Kesteven are currently subject to algorithm-assisted decisions. SKDC's ICT Strategy 2025–28 commits to incorporating AI, machine learning, and automation into service delivery, but it does not specify where those tools have been or will be applied. That absence of specificity is not unusual — many councils are in the same position — but it has a practical consequence: a resident in Stamford or Market Deeping currently has no public document telling them whether an algorithm contributed to a decision made about them, or what form that contribution took.
The risk SKDC already named
The council has already named the problem. SKDC's ICT Strategy 2025–28 embeds an AI SWOT analysis in its Strategic Risk Register and formally identifies 'AI governance lag' — the risk that reliance on AI outpaces the checks and balances needed to govern it responsibly — as something its committees must actively monitor. That is not a trivial admission. Placing a governance failure mode on a risk register makes it visible to elected members and senior officers, not buried in a supplier contract or an IT team's internal notes.
But there is a difference between institutional self-awareness and public accountability that matters in practice. A resident affected by an algorithm-assisted housing or benefits decision does not benefit from a line in a risk register they cannot read and would not know to look for. The Strategic Risk Register is a governance instrument for the council. It is not a transparency mechanism for residents.
As of the research date for this article, South Kesteven does not appear in the GOV.UK Algorithmic Transparency Recording Standard (ATRS) repository — the national register where public bodies publish records of their algorithmic tools, including how those tools work and what safeguards apply. That position may change; local authorities can submit ATRS records voluntarily, and some have begun doing so. But no such record for SKDC had been identified at time of writing.
The council has taken the first step: recognising that AI adoption creates a governance risk. What the public record does not yet show is whether that recognition has produced resident-facing measures — a way for someone in Bourne or Stamford to understand how an automated decision affecting them was reached, or to ask for human review.
What national frameworks cover — and what they don't
The structural gap is not unique to South Kesteven — it runs through the architecture of UK algorithmic accountability itself.
The Algorithmic Transparency Recording Standard was launched by the Government Digital Service in 2021 and made mandatory across central government departments and arm's-length bodies in 2024. By May 2025 its repository held 59 published records. Only 10 came from devolved administrations and local government combined. Local councils are strongly recommended to comply but are not legally required to do so. That distinction — advisory for the tier of government closest to most people's daily lives — is where the accountability gap sits.
A Lords bill introduced in December 2024 is pushing to close it. The Public Authority Algorithmic and Automated Decision-Making Systems Bill would require impact assessments before public authorities deploy AI and add explicit protections where automated decisions carry legal or significant effects on individuals. It has not yet passed.
The moral case for urgency is made clearly in the LSE Public Policy Review: unlike a decision by Amazon or a bank, a council decision on housing, benefits eligibility, or legal status cannot be declined or taken elsewhere. There is no opt-out from the state. That coercive quality — which makes public services valuable — also makes algorithmic errors in local government more consequential than equivalent errors in the private sector.
Civil society is filling part of the gap in the meantime. The Public Law Project's Tracking Automated Government register had recorded 55 algorithmic tools in administrative decision-making by 2025, and a Nuffield Foundation-funded project is mapping what information residents actually need to mount a legal challenge when something goes wrong.
The LGA's survey conducted over December 2024 to February 2025 found that the vast majority of English councils were already exploring or using AI. South Kesteven is not an outlier — which is precisely why the absence of a mandatory local accountability framework is a systemic issue rather than a local failing.
The extra layer: rural exclusion and digital access
Rural geography sharpens everything. Around 12% of rural UK homes lack access to superfast broadband, and roughly 21% of people nationally lack the fundamental digital skills needed for everyday life — figures that are likely higher in Lincolnshire, where older demographics and dispersed village communities compound both barriers. South Kesteven sits inside that pattern: the towns of Grantham, Bourne, Stamford, and Market Deeping each have surrounding villages where connectivity and digital confidence vary considerably.
The double disadvantage is specific. Residents without reliable internet or digital literacy may be disproportionately affected by automated council decisions — on housing waiting lists, council tax reduction, or eligibility assessments — precisely because digital-first service design increasingly assumes a connected, confident user. At the same time, those residents are the least equipped to identify that an automated decision was made about them, understand what it involved, or navigate an appeals process that presupposes digital access.
Lincolnshire's Integrated Care System Digital Inclusion Strategy 2025–28 acknowledges this tension directly. It commits to retaining face-to-face access alongside digital channels and explicitly covers groups often overlooked in digital transition plans: GRT communities, seasonal residents, and older people lacking digital confidence. That commitment matters, and it is worth being clear about its limits: preserving access to services is not the same as preserving access to accountability. Knowing a decision has been made about you, understanding how, and knowing what to do about it requires a layer of transparency that connectivity alone cannot supply.
What accountability could reasonably look like
Three things would make algorithmic accountability meaningful at district level, and none requires waiting for Parliament. A council would need to publish which service decisions involve algorithm-assisted tools; offer plain-language explanations available to any affected resident; and provide a clear route to human review. That is not a high bar — it is closer to good administrative practice than to a novel technical challenge.
For a resident in Bourne or one of the villages east of Grantham, one concrete right already exists, even if it rarely appears in council communications. Under UK GDPR Article 22, anyone subject to a solely automated decision that produces a significant legal effect has the right to request human review of that decision. That right sits inside existing data protection law. It is available today, without any new legislation.
Voluntary compliance with the Algorithmic Transparency Recording Standard — which local councils can submit to without a legal mandate — would not of itself give that resident any remedy. A published record explains what a tool does; it does not reverse a wrong outcome. But it would at least mean that a person in Stamford questioning a council tax reduction decision, or a family in Market Deeping challenging their position on a housing waiting list, could find out whether an algorithm played a role in the first place.
The question posed at the start of this article — who is responsible when the algorithm gets it wrong? — does not have a tidy answer for South Kesteven's residents right now. The council has put the governance risk on its own register. What the public record does not yet show is whether that internal recognition has produced anything a resident in Grantham could actually use.
- [1] South Kesteven – Wikipedia. https://en.wikipedia.org/?curid=426477 https://en.wikipedia.org/?curid=426477
- [2] Government by algorithm – Wikipedia. https://en.wikipedia.org/?curid=63424816 https://en.wikipedia.org/?curid=63424816