AI and Housing: Re-imagining the future of service delivery

The AI hype / reality check

“AI could create an immortal dictator, from which there is no escape”, warned Elon Musk in a documentary released last year. His concern is that if one company or a Government can develop super intelligent Artificial Intelligence (AI), then that super intelligence could outsmart humans and become our master. This is the branch of AI referred to as Artificial Super Intelligence – defined by Oxford philosopher Nick Bostrom as “Any intellect that greatly exceeds the cognitive performance of humans in virtually all domains of interest. You’d like to think Elon Musk knows what he is talking about given his investments in AI through his companies such as OpenAI, Neuralink and Tesla.

But futurist Ray Kurzweil advises “AI Will Not Displace Humans, It’s Going to Enhance Us”. He envisages a future world (2045) where humans will link wirelessly from our neocortex - the portion of the cerebral cortex that serves as the centre of human higher mental functions - to a synthetic neocortex in the cloud. We will connect with machines via the cloud and we will also be able to connect to another person’s neocortex. Essentially, a merged world of human and computer (machine) coexistence. He believes this could enhance the overall human experience and allow us to discover various unexplored aspects of humanity.

It is amazing to think that a term originally created in 1951 by Prof John McCarty, Professor at Stanford University could 68 years later potentially have such a seismic shift on our future world.

How does that future world relate to where we are today with AI and the endless hype and talk about robots replacing our jobs?

In their analysis of the future of Work, McKinsey have predicted that up to 30% of the world’s jobs could be automated by 2030 and half of the work we do today could be automated with current technologies. Whatever the numbers, automation and the application of AI in the workplace will impact us. It is already happening and we need to be prepared. Technology displacing people and people and technology working together is nothing new. It has been a constant evolution. With AI, the difference is that the rate and level of impact is more profound. We will need to develop the skills and knowledge to work with and not be displaced by AI. However, we need to be clear about the scope of AI that is becoming increasing influential.

The reality is that Artificial Super Intelligence (as described above) and even Artificial General intelligence (what you see with Tony Stark in Iron Man) is some way off. For machines to achieve true human-like intelligence, they will need to be capable of experiencing consciousness, empathy and emotion. Unlikely - I think! However, we will certainly see Narrow AI (Amazon Echo, IBM Watson and Google Duplex and others) develop further. Narrow AI systems are able to process data and complete tasks at a significantly quicker pace than humans, allowing us to improve our overall productivity, efficiency, and hopefully quality of life.

For many housing associations, just when you’re going digital and cloud, what will this narrow AI mean for your organisation? When some experts are now saying you’ll need a Chief AI Officer, never mind a Chief Digital Officer, will you be AI ready when you might still be facing the digital ready challenge?

AI and Machine Learning 101

In his article in DIN bulletin 1 (November 2018), Arturo Dell from HouseMark offered the definition of AI from Russell and Norvig – “AI is the designing and building of intelligent agents that receive precepts from the environment and take actions that affect that environment”.

For me, Artificial Intelligence is simply the science of getting computers to learn and act like humans do, and improve their learning over time in an autonomous fashion. They achieve this by being fed with lots of data and other information in the form of observations and real-world interactions.

The key developments we are seeing in AI are in the areas of Machine Learning and Deep Learning.

Machine Learning is a subset of AI which uses statistical methods to enable computers to improve with experience gained from learning. Statistical mathematical methods have been around a long time. The difference is we have now combined the processing and storage power with the larger data sets to enable improved learning and hence reliable application of machine learning algorithms. That’s one big reason why Google loves to collate your data.

There are many different types of machine learning algorithms, with hundreds published each day, and they’re typically grouped by either learning style (i.e. supervised learning, unsupervised learning, semi-supervised learning) or by similarity in form or function (i.e. classification, regression, decision tree, clustering, deep learning, etc.).

Deep Learning is a subset of AI which makes the computation of multi-layer neural networks feasible, more closely mimicking how our brain works. It is different from traditional machine learning in that it looks for features or patterns in the data from which it can improve its learning. This is in the land of Google Deep Mind and the technology behind Google Duplex. Deep Learning has now risen from the world of academia research into the commercial world and is the root cause for much of the hype today. It is the area of AI that will drive (literally) autonomous self-driving cars. It is where and how computers are achieving near accurate natural language speech recognition through the use of recurrent neural networks. But don’t be fooled into thinking it is a panacea for how AI will solve any problem. Behind the Google Duplex appointment booking example lies an army of employees behind the scenes manually defining intent models, training them with relevant utterances, and then connecting them to hand-authored responses.

The potential of AI for housing

I recently presented and hosted a panel discussion on AI at Tech@Housing where two key areas for the application of AI were particularly highlighted:

·        Optimising customer experience

·        Improving maintenance – making it more predictive and pre-emptive

Customer experience examples include:

·        The infamous chat-bot - Used to automate customer service text conversations

o  The rise of chat-bots has been driven through people using mobile devices more and more rather than the traditional keyboard/mouse/monitor interfaces. It is easier to use a chat interface then expect a user to download yet another app onto their device

o  Unexpected questions will likely ‘break’ the chat-bot system so consumers need to be clear that they are interacting with a machine

o  KLM automates responses to over 50% of customer enquiries on social media by implementing a machine learning chat-bot. TicketMaster uses conversational voice and text chat-bots to improve event searching and ticket sales experience. Booking.com now resolves half of customer queries to its text chat-bot in five minutes and without human intervention using semi-supervised learning. Marks & Spencer plans to automate all customer call routing with 90% accuracy using machine learning.

·        Analyse and understand customer sentiment displayed through direct customer contact:

o  Using voice and text analysis to uncover overall customer sentiment – negative or positive – sometimes in real-time as displayed when they contact the company, for example through a contact centre

·        Monitor customer experience across multiple channels to build a holistic overview and identify high profile and high priority issues:

o  Advanced analytics on all customer contact data across multiple channels to uncover insights to improve customer satisfaction and build a holistic picture of their status

Maintenance examples include:

·        Automate inventory management for spares

o  Calculate and predict how many units are required where and when, in order to reduce inventory costs and minimise obsolete and excessive inventory

·        Minimise the costs for replacement and/or upgrading of failing or under-performing parts or products. Rolls Royce plans to predict maintenance requirements for jet engines to improve aircraft efficiency using Microsoft Azure's machine learning

·        Improve preventative maintenance and Maintenance, Repair and Overhaul (MRO) performance with greater predictive accuracy to the component and part-level. Predictive maintenance predicts when certain products or devices are in need of maintenance what sort of maintenance, the likely maintenance and replacement materials, and technician skill sets. EDF Energy plans to monitor power station conditions in real time and predict maintenance requirements using machine learning

·        Optimise job scheduling based on factors such as weather, estimated travel times, technical capabilities and parts availability

·        Optimise staff transportation routing based on factors such as weather, traffic, changing job loads and shifting priorities

While many housing organisations are still focused on digital as the main enabler for transforming services, there are benefits that machine learning will clearly bring. This will invariably be combined with other emerging and existing technologies and of course relies on quality data with volume for the algorithms to work effectively to give valuable business insight. That and a good dose of cultural change management will be crucial for success, with respect for data privacy. We can re-imagine the future of service delivery in this way. And it won’t be about the robots taking over but true social landlords delivering a better valued and for value service to customers through innovative and appropriate use of just another emerging technology.