Housing Digital take a brief excursion into the ways AI is shifting the social housing landscape – but it’s a journey not without its potholes
Across the UK, social landlords are quietly experimenting with a new set of tools that are said to promise a revolution in housing management.
The potential of this technology is said to be wide-ranging, able to turn reactive repairs into predictive upkeep, spreadsheets into insight, and faceless contact centres into personalised support; clearly, there’s already a great deal of optimism invested in artificial intelligence (AI).
From humidity sensors that spot the early signs of mould to algorithms that help prioritise scarce maintenance resources, social landlords are exploring how AI can be folded into their everyday business of keeping homes safe and affordable – and, crucially, improving tenants’ lives.
From drip to dashboard
One of the clearest early wins is predictive maintenance.
By combining building data, tenant reports and historical repair logs with machine learning, housing providers can flag likely failures before a boiler breaks down or a lift fails – reducing costs and disruption, and improving first-time fix rates.
Vendors and systems used by councils and associations report reductions in downtime and inventory costs when predictive analytics are applied to plant and asset management.
Environmental monitoring is another rapidly scaling area. Firms such as Switchee supply smart sensors to social landlords to track temperature, humidity and pressure in flats. The data these devices collect helps spot the conditions that cause dangerous mould and damp – problems that have tragically been linked to serious illness and even death in recent years.
Landlords using such tech say it lets them move from a reactive approach to a preventative one, intervening earlier and targeting support to vulnerable households.
AI also sits behind modern tenant engagement platforms. Chatbots, automated triage systems, and appointment-scheduling algorithms are said to reduce the administrative burden on staff and give tenants quicker answers outside office hours.
Central government initiatives are likewise experimenting with assistant-style AI to cut admin time in local services – a sign that digital assistants could become a mainstream channel for housing advice and casework.
Beyond operations, AI is being used for strategic problems: data models that map which neighbourhoods need retrofit investment, optimisation tools that reduce void times, and even predictive analytics to identify households at higher risk of homelessness so early support can be offered.
Those applications are still novel, but they point to an ‘intelligence’ that augments hard decisions about scarce public funds.
Why landlords are attracted – and cautious
The attraction is obvious: better targeting, lower costs and faster service.
Early adopters report improvements in efficiency and tenant satisfaction when technology is paired with clear operational changes. But the sector’s move to AI also exposes it to significant risks: algorithmic unfairness, data protection failures, supplier concentration, digital exclusion and reputational damage if things go wrong.
The technology – in some cases the technology providers – can provoke controversy, too, making caution a must. As reported in The Guardian recently, Coventry City Council’s decision to use Palantir AI to assist in the delivery of some of its services has provoked some disquiet (to put it mildly).
If nothing else, it shows how quickly trust can evaporate if landlords do not handle procurement and communication carefully.
Any new technology brings issues and challenges, and AI is certainly no different. Some of the main areas of concern include:
- Data protection and privacy: Social landlords routinely process sensitive personal data. AI projects that combine tenancy records, health indicators, or social-care notes multiply the risk profile. Landlords must carry out data protection impact assessments (DPIAs), apply lawful bases for processing, and minimise what they collect. It’s advised that analytics are kept on pseudonymised datasets where possible, and ensure clear retention and deletion policies. Engage tenants early is further advice – explain what data is used and why, and offer opt-outs where feasible
- Bias and fairness: Algorithms trained on historical records will mirror the inequalities in those records. If risk-scoring tools are used to prioritise services, they can entrench disadvantage unless audited. Advice is to run fairness checks, use diverse training data, and retain human oversight for decisions that materially affect tenancies or access to support. Independent algorithmic audits are considered best practice
- Digital exclusion: Not all tenants can or want to engage through apps or chatbots. Technology must augment, not replace, human channels. Telephone and in-person routes remain a desirable option among tenants, so it remains important to ensure frontline staff are trained to help digitally excluded residents
- Governance, skills and accountability: Boards also need to understand AI’s limits. One response to this issue is the appointment of so-called AI or data champion, who can help to map out the line of accountability, and build simple governance around model deployment, monitoring, and incident response
AI technology is still relatively new, so there’s no set route to implementation; everyone’s working it out on the go, as it were – even the big tech firms behind it (one suspects).
As a generalised rule of thumb, it’s considered a sound move to start small, with pilots focused on clearly measurable outcomes (such as void reduction, mean time to repair, mould interventions). It’s advisable to keep humans in the loop for consequential decisions.
Adopt privacy-by-design, is another important consideration, as is the publishing of plain-English model descriptions, and also the use of third-party audits for fairness and security. Other considerations are to prioritise tenant engagement: test interfaces with residents, co-design consent flows and publish impact assessments. Finally, it’s important to share learning across the sector.
Better homes, not surveillance
AI is said to promise efficiency and more personalised support for tenants, but with it also comes the fears of landlord over-reach – of ‘prying eyes’ and stern micro-management.
It’s fair to say that such concerns are not entirely unfounded (an issue applicable far beyond the social sector, of course). Like many a technology, it can be used to serve – or it can be used to control.
Properly governed, it could help landlords find damp before it makes people ill, get repairs done first time, and spot households heading towards crisis so interventions can be made earlier. But the same tools can feel intrusive if deployed without explanation, or unjust if they automate decisions about people’s homes.
The test for the sector will be simple: can landlords use data to improve lives while keeping tenants – and their rights – at the centre of design?
Overall, the social housing sector is showing it is neither technologically naive nor blindly enthusiastic. That’s perhaps just as well, because for all the genuine potential in the technology, much hype abounds that would have us believe it’s the miracle silver bullet to everything.
As ever, it’s about learning to pair old-fashioned human judgement with new-fangled prediction. Those who stitch AI into their operations thoughtfully – with strong data governance, clear procurement, tenant engagement and an ethic of augmentation rather than replacement – will probably unlock meaningful gains for tenants.
Those who rush may quickly find that public trust is far harder to repair than a broken boiler.
Image credit: Jirsak/Shutterstock
Read next: Emergency action plan is a start, say Scotland’s housing leaders
Are you a social housing professional? Sign up for a FREE MEMBERSHIP to upload news stories, post job vacancies, and connect with colleagues on our secure social feed.
