Bad data is a big issue for AI, so where does that leave social purpose?

AI is fast becoming central to every business strategy, but without the data to back it up, housing providers will find themselves unable to deliver on their social purpose, argues Trevor Hampton

MANY housing providers have signed up to the universal belief that artificial intelligence will help them manage the dilemma of rising demand versus diminishing resources.

Trevor Hampton
Trevor Hampton, director of housing, Northgate Public Services

And the stakes are high. An aging population, a rise in homelessness and dealing with the complex needs of many tenants, means it’s more vital than ever for providers to know how best to target and spend the funding available.

But whilst it’s true AI will help identify where best to meet those needs; the real game changer is data. Few sectors are as ‘rich’ in data as housing, but to truly benefit from it, the groundwork needs to be laid now.

Here are three steps you can take to ensure AI will be as transformational as the hype suggests.

1: Embed data in your DNA

AI’s ability to cross reference and spot patterns will deliver untold benefits for the social housing sector, but its success will not be down to technology alone.

The technology is just one part of a triangle, that needs the right people and the right processes in place to capture the real powerhouse behind AI and data.

Gaps in the data will lessen the opportunity to form a holistic picture of tenants, meaning the AI will return rudimentary answers at best.

How can you avoid the pitfall of incomplete data sets? The answer is to take your staff on the journey with you. If no one has explained the importance of scanning the letter, logging the call, or recording the repair in the Housing Management System, then it could well get overlooked. The detrimental effect will be insufficient information to train the algorithm later down the line.

Fostering a whole organisation approach will help encourage staff to think how AI could be used to make them not only better at what they already do, but importantly, what data needs to be captured to train it.

Poor results happen when there is a lack of understanding, so create a workplace culture that appreciates how data can be used to improve outcomes for both themselves and tenants. Automation of routine tasks will free them to upskill and take on more varied and challenging roles and customer service will be improved, but not if the data is missing.

2: Be data savvy

Now is the time to interrogate your data, if it housed in multiple systems, then it will create different answers to the same questions.

Good quality data is essential, as duplication increases inaccuracy, whist incomplete data, will mean pieces of the puzzle are missing. As a consequence, when the time comes to apply AI, the opportunity to drive efficiency savings or take early intervention, will be lost.

Take a quick works order raised for a repair call out and the lack of recorded information as a result. Was it a one-off repair? How old was the item? Was the call out to a boiler or a leaky shower? It might be a quick fix now, but later on without sufficient detail, AI can’t be applied.

This is because if there isn’t a record of the repair, then any subsequent analysis will be restricted and the opportunity to predict when a repair should be carried out or if a replacement should be ordered will be lost. Translating into unnecessary call outs, which can not only harm the tenant/landlord relationship but is an inefficient use of resources.

Times the repair by 20 and you get the picture.

The right data will enable AI to identify if all the original work was by one contractor or the items were from one manufacturer. Being able to apply the learning across the whole database, would enable the identification of similar properties likely to be affected. This would then allow for a replacement schedule to be drawn up, to tie in with scheduled maintenance, reducing the need for unnecessary visits and repairs. A win, win.

To maximise the potential of AI it’s important to think beyond simply going digital. Implementing a 24/7 self-service option accessible from anywhere at any time, will not be enough to ensure there’s sufficient data to train the AI in three years’ time. To achieve this, it’s imperative to take steps now to scan and upload a digital record of all the previous correspondence. If it remains in a filing cabinet by default, then the AI won’t deliver what you expect, building accurate data sets now will pay dividends later.

To provide actionable insights into tenant behaviour there needs to be machine learning across the whole database, built up over time.

3: Asset Management

Bill Gates has described AI as the ‘Holy Grail,’ but for it to prove as transformational as the buzz around it suggests, there needs to be a recasting of the role of data from operational concern, to key business asset.

The reality is many housing providers are failing to grasp the importance of laying the data foundations needed for AI to make a tangible impact on the overarching business strategy.

We undoubtedly live in an information age. In fact, the velocity is such, that without a proper data vision, organisations could find their ability to capture it, outstripped by the pace by which it’s growing.

Failure to put data first and foremost, could leave housing providers looking to turbocharge their decision making through AI, at best underwhelmed and at worst, unable to fulfil their social purpose.

The takeaway? AI may be the rising star, but without the right data to back it up, you could be disappointed by its performance.

NH

Trevor Hampton is director of housing at Northgate Public Services.

 

%d bloggers like this: