Civica’s Alex Oldman explores how the internet of things is already making a big difference in social housing, but there’s still some way to go.
Artificial intelligence (AI) is proven to deliver real benefits and protect residents from harm using pattern cognition and augmented reality. While the UK government announces plans to introduce ‘Awaab’s law’ into the forthcoming Regulation of Social Housing bill, technology already exists to protect vulnerable households from hazards such as damp and mould. But why hasn’t this technology been widely deployed?
There are examples of early adopters. Flagship Housing is two years into its programme to deploy 20,000 Switchee devices to homes across the east of England by 2030. These devices monitor home environments, harvesting data and looking for patterns. The software then adjusts the heating system, saving the household money and protecting the building and residents against risk of damp and mould. Pre-emptive boiler-breakdown alerting also improves Flagship’s repairs service and customer satisfaction. AI learning can be used to determine the deployment priorities based on occupant need, built form and other human-determined factors.
Kingdom Housing has implemented visual AI in the form of augmented reality (AR) which overlays digital information onto the real world. The organisation deployed an AR solution during the pandemic to provide remote guidance for responsive repairs. The solution allows repair operatives to provide a hands-on overlay within an app to show how to resolve a problem.
Visual AI gives software the skill to scan, identify and classify objects from video or still-image sources. This allows solutions that can recognise and understand an image they are being presented with. For example, using visual AI to inspect photos provided by residents for evidence of damp can provide protection or drone surveys can identify problems with roofing that might be invisible to the naked eye, plus a drone can survey hundreds of houses per day from the air, saving time and efficiencies.
Meanwhile, text-based AI enriches customers’ digital experiences by allowing searches based on semantic analysis and natural language processing (NLP) to improve interaction with searches and chatbots.
Functional AI can detect a problem by comparing an actual working pattern with a standard one. A good example of this is plant-room monitoring, where the acoustic signature is first learned and then anomalies are found by algorithms to find faults in, say, bearings. Preventative maintenance can then be scheduled to prevent downtime from unplanned breakdowns.
It is important that organisations must be very clear on their policies before deploying any type of AI. As the writer Terry Pratchett once said, “Real stupidity beats artificial intelligence every time.”
Huge data capture and data processing activities must be balanced with data privacy. For example, automated decision-making must be aware of any hidden or unwitting bias that might not consider race, age or gender. But with all these considerations taken into account, it’s clear that change is coming, and we need to be ready for it.