Decisions happening before you arrive
Picture a Tuesday morning in Grantham. Someone rings their GP surgery to rebook a hospital appointment. The receptionist offers a slot, the call ends in two minutes, and life carries on. What neither of them knows is that the time offered may already have been shaped by software: an algorithm that has weighed anonymised patient records, local employment patterns, and perhaps even the weather forecast to predict which slots are most likely to be kept.
This is not science fiction, nor a warning about some future system. Across English NHS trusts, tools of this kind are already running. South Kesteven residents — whether they are navigating a council service, applying for a job through an online portal, or simply ringing for an appointment — are increasingly encountering decisions that were partly formed before they picked up the phone.
Call them quiet algorithms: automated processes that are real, consequential, and largely invisible to the people they affect. The purpose of this piece is not to argue that these tools are harmful. Many show genuine promise. The question is a simpler and more democratic one: if algorithms are already embedded in the services that people in South Kesteven rely on, what do local residents actually know about them?
The NHS appointment algorithm and how far it has spread
Deep Medical's software, commissioned by NHS England and expanded in March 2024, does something that sounds almost mundane: it tries to predict which patients will miss their appointments. To do this, it draws on anonymised patient records alongside external signals — weather conditions, traffic data, local employment patterns — and uses that blend to flag slots most at risk of being wasted. At Mid and South Essex NHS Foundation Trust, where a six-month pilot ran before the wider roll-out, the results were striking: a 30% fall in missed appointments, 377 DNAs prevented, and an estimated saving of £27.5 million a year. NHS England subsequently announced the system would extend to ten more trusts across England.
United Lincolnshire Hospitals Trust, which serves much of South Kesteven, is not among the confirmed sites. There is no public evidence placing ULHT inside this specific programme. That is worth stating plainly — but it is also, itself, the point. With the national expansion ongoing and NHS Digital explicitly backing AI scheduling tools as part of broader efficiency drives, the question of whether such software reaches Lincolnshire is less 'if' than 'when'.
In the meantime, residents booking through any NHS trust have no straightforward way to find out whether scheduling logic of this kind is shaping their care. The algorithm, if present, is not announced at the point of contact. There is no opt-out, and no plain-language explanation routinely offered. That is a transparency gap before it is anything else.
Who gets left out when NHS services go digital
The efficiency case for digital NHS services is straightforward; the equity case is more complicated, particularly outside cities. South Kesteven is a district of market towns — Grantham, Stamford, Bourne, the Deepings — surrounded by villages and hamlets where connectivity is patchy and population age skews older. The national figures give that geography some texture. NHS England's own 2023 digital inclusion framework recorded around 10 million adults in England without foundation-level digital skills, 7% of households without home internet, and — most pointedly — roughly 30% of people who are offline describing the NHS as one of the hardest organisations for them to deal with. Not one of the easiest: the hardest.
The NHS Confederation's 2026 digital exclusion index sharpens the local concern further. Rural and coastal areas, precisely the landscape that defines much of South Kesteven beyond the A1 corridor, score consistently higher on exclusion. The government's stated ambition to make the NHS App the universal 'front door' to care by 2028 is therefore not an abstract policy goal for districts like this one — it is a live equity question.
None of this means digital transformation is wrong. Faster, smarter scheduling genuinely helps patients who can use it. But the same systems that save time for one resident may quietly shift the burden onto another. Keeping telephone and face-to-face routes open is not sentiment; in a rural district with an ageing population and variable broadband, it is a practical prerequisite for care that reaches everyone. That is a political and institutional choice, not a technical default.
South Kesteven council's AI and the transparency gap
The council itself is not standing still. South Kesteven District Council's digital strategy is framed around intelligent automation and predictive analytics — forecasting, for instance, peaks in customer service demand so resources can be positioned in advance. Its official position is that high-stakes decisions on planning, legal matters, and welfare remain with human officers and elected committees under the council's Overview and Scrutiny framework. SKDC's membership of the LGA's Advanced and Predictive Analytics Network, whose resources include guidance on 'explaining decisions made with AI', suggests this is active engagement rather than passing interest.
LGA surveys conducted between December 2024 and February 2025 found approximately 95% of responding English councils already using or exploring AI, mostly predictive and generative tools. South Kesteven is almost certainly somewhere in that figure. But here is the governance problem: unlike central government departments, for which the Algorithmic Transparency Recording Standard became mandatory in 2024, local councils face no equivalent requirement. The entire national ATRS repository contains only ten records from devolved administrations and local authorities combined, out of 59 published in total. That means the algorithmic activity of most English councils — including, as far as public records show, South Kesteven's — is simply invisible to residents.
This is not a South Kesteven-specific failure. It is a structural gap affecting almost every local authority in England. The tools for transparency already exist; ATRS is a working framework with both plain-English and technical tiers. What has not happened is any political decision to extend its reach to local government. Until that changes, residents have no practical mechanism to know what is automated, what data it uses, or how to challenge an outcome it shapes.
What AI job screening does to rural applicants
Employment is where algorithmic sorting reaches furthest into ordinary life. Around 99% of large employers now use some form of automated screening at an early stage of recruitment — a figure that means any South Kesteven resident applying to a major retailer, logistics operator, or public-sector body is likely to encounter one before a human being sees their name.
The risk is not that these tools are universally biased. It is that they are structurally predisposed to disadvantage certain profiles. Research from Newcastle University and Stanford's Human-Centred AI institute has documented how proxy discrimination operates in practice: postcodes that correlate with lower socioeconomic status, school types associated with particular regions, career gaps, or non-linear work histories can all silently suppress a candidate's score without any explicitly discriminatory instruction ever being written. South Kesteven's mixed rural economy — agricultural workers, part-time and seasonal employees, carers returning to work, people whose working lives have tracked the rhythms of land rather than career ladders — fits that at-risk profile closely.
The UK government's March 2024 Responsible AI in Recruitment guidance, developed with the ICO, the Equality and Human Rights Commission, and the Ada Lovelace Institute, names discriminatory targeting and digital exclusion as genuine harms, not hypothetical ones. Crucially, it confirms that employers cannot transfer legal liability to the vendor whose software they deploy — under the Equality Act 2010 and UK GDPR, accountability remains with the organisation making the hiring decision. That is a meaningful protection, but only if the candidate knows what shaped the outcome in the first place.
What trust actually requires when decisions are automated
Trust is the word that tends to close these conversations, and it deserves more precision than it usually gets. A 2022 academic review of AI in public services found that algorithmic systems do not simply replace human judgement — they redistribute it across software, vendors, and officials in ways that make it genuinely difficult to identify who is responsible when something goes wrong. The chain of accountability becomes, in the researchers' phrase, modified rather than maintained. That is not an abstraction: it describes what happens when a resident in Grantham cannot get a straight answer about why a council service flagged their case, or why a job application stalled.
Challenging an automated decision is harder still. A 2024 study on contestability in public services found that meaningful redress typically requires an NGO worker or lawyer — someone with the knowledge and leverage to navigate a system not designed to be questioned. That is not a realistic option for most people. Subject Access Requests under UK GDPR are a genuine right and worth using, but they reveal what data an organisation holds, not always how the algorithm used it. They are a starting point, not a solution.
What does change things, according to a 2025 experimental study of 4,087 participants, is transparency and visible human oversight — knowing that a decision was automated, understanding the broad logic, and seeing that a named person can intervene. There is a telling detail in those findings: administrators showed significantly greater readiness to accept automated decisions than citizens did. That gap — between institutional comfort and public trust — is exactly where governance either earns legitimacy or loses it.
For South Kesteven, good algorithmic governance would look like something straightforward: residents able to ask whether a decision involved automated tools and receive a clear answer; a published record of what those tools are; and a named human route for challenge. None of that requires dismantling the efficiency gains. It requires treating legibility as a design requirement, not an afterthought.
- [1] The loopholes of algorithmic public services: an 'intelligent' accountability research agenda. (2022). https://doi.org/10.1108/aaaj-06-2022-5856 https://doi.org/10.1108/aaaj-06-2022-5856
- [2] Algorithmic Governance: Experimental Evidence on Citizens' and Public Administrators' Legitimacy Perceptions of Automated Decision-Making. (2025). https://doi.org/10.1111/padm.70028 https://doi.org/10.1111/padm.70028
- [3] Understanding Contestability on the Margins: Implications for the Design of Algorithmic Decision-making in Public Services. (2024). https://doi.org/10.1145/3613904.3641898 https://doi.org/10.1145/3613904.3641898