AI and Real Estate

0 (2).jpeg

My presentation, delivered at the Future: PropTech Conference in London on the 2nd May, 2018. 15 minutes.

Or if you prefer to read it .......

Ok, so we are going to talk about Use Cases for AI

The what, the so what, and the now what

To start off I’d like to show you what the 1st industrial revolution felt like to Victorians. This is Rain Steam Speed by Turner, painted in 1844, just after the opening of the Great Western Railway.

Steam engines transformed Victorian society as dramatically as AI is going to transform our society. The future feels like it is coming at us fast and everything is a bit of a blur.

But the excitement is palpable - we are lucky to be living in such fast moving times.

And fast moving they really are.

In January last year McKinsey wrote that 49 percent of all activities people are paid for could be automated by currently demonstrated technology.

Not technology from the future, technology that is available today.

The RICS reported that 88% of the core tasks performed by surveyors could be automated over a ten year period.

So... what is happening to generate all this change?

Well the answer starts here. This is Moore’s Law, where essentially computing power doubles every two years, something which has held true for 50 years so far.

So you have 100 thousand transistors on a chip in 1980, 100 million by 2000 and 10 billion by 2016.

This is the archetypal example of exponential growth, the famous hockey stick we hear so much about.

But, compared to the chips that originated in gaming machines but that are now the primary processing units behind AI, this Moore’s Law growth is rather paltry.

Moore’s Law is shown in the two aneamic lines at the bottom. The power of GPU’s has been increasing dramatically faster over the last few years.

Which is why neural network training, which is fundamental to machine learning, became 60 times faster in the three years from 2013 to 2016, with most of that growth in one wonder year, 2015. 

So what does this lead to then?

Well for one thing it means computer vision, face recognition etc has gone from being useless to a utility in just a few years.

With error rates of 28% in 2010 it was a pretty useless technology. By 2017, the error rate was down to 2.2%, far better than the 5% which is how humans perform.

And unlike humans, who can only process a few images at a time, computer vision enables 200,000 faces to be recognised in real time.

Likewise speech recognition has gone from useless to utility. In fact it has done so considerably faster than computer vision.

In 2013, we saw word/error rates of 23% - just four years later this was below 5%, again the human level of achievement.

Which is why we have seen all of a sudden the rise of smart speakers like Alexa, and the growing use of Voice as a search interface

It is what happens when new technologies become utilities

So, ALL businesses can now exploit 6 new technological capabilities, made possible by the growing power of AI.

Many more processes can now be automated

We can understand what is happening in pictures and videos

We can optimise complex systems in a manner that was not possible before

And we can automatically generate voice and textual content

And we can automatically understand people using language and make predictions.

Within the real estate industry we have mapped these new capabilities as applying across 17 areas or workflows.

Much of the day to day requirements of the industry can be improved using AI 

From investment strategy, to asset monitoring, to customer experience to demand optimisation. The possibilities are almost endless.

But, with AI, there is a catch.

Unlike previous technologies, where often the best policy was to let early adopters make all the mistakes and lose a lot of money, with AI you cannot be a fast follower, and ride on the coat-tails of innovators.

You have to be the innovator

Because AI is all about data and the training of systems. Machine learning means just that - the machines learn from experience and that is a step you cannot buy your way around.

Experience matters in AI. And the rewards go to those who have learnt the most. 

So, what then do we do about all of this? How do we apply these powers to real estate?

Well, across real estate there are a number of key things we all have to do.

But really everything boils down to making predictions, and AI is dramatically reducing the cost of prediction. Who will buy, who will sell, what happens if this, what happens if that. What are the chances of etc etc etc

And with better and cheaper predictions, the amount of uncertainty we have to deal with reduces. And less uncertainty is a great enabler. We can take more risks, at less risk.

Here are 9 key Use Case areas for AI in real estate

Computer vision is a powerful weapon. Being able to point a camera at the built environment and to understand what is being seen is a superpower.

But people also like interacting with images. They are comfortable with images. Real estate search using images rather than text will become a big thing.

The internet of things and smart buildings and smart cities, are going to allow us to optimise ‘the space around us’ in unprecedented ways.

Not least of all we are going to be able to optimise our workplaces so that they at last allow us to be as productive as we can be.

All around the world there are now thousands upon thousands of small low altitude satellites and these are allowing us to monitor, inspect and analyse the world in great detail.

We can merge satellite and aerial imagery with other data sets to aid our understanding. From daily monitoring of development sites to checking how full car parks are at shopping centres, there is a lot you can do when you can analyse imagery.

AI is extremely useful in synthesising disparate data sources and finding correlations and causations between them. Automated Valuation Models will become more and more accurate.

And a sidenote on that: As we have said AI can be a winner takes all technology - as with web search we probably only need one automated valuation model in each asset class.

Chatbots are a key AI tool. For anything deeply structured allowing people to text for help or assistance makes a lot of sense. We’ll see a lot of chatbots in property management and customer support.

Location, location, location is of course a real estate mantra but historically we have not really known that much about any given location. AI will allow us to become far better informed about the precise characteristics of locations, and the wants and needs of the people who live, work or shop there. 

Voice as a search interface is now becoming commonplace with the likes of Alexa. Expect to see a lot more people talking to computers rather than tapping away at keyboards.

Natural language processing is all about understanding text, and once that becomes a solved problem, all manner of interactions we currently have with text will be changed. Expect to see a rapid decline in human input into report creation for example. Certainly in a business context most reports can be boiled down to a templated format, and auto generated by linking in various data sources.

And finally marketing. Probably the No1 use case for AI across all businesses. Marketing is all about knowing your customer and AI, in various ways, is going to enable us to be much much better at that. Real estate has a lot to learn here. Those who bother to learn will reap big advantages.

So, nine areas where you will find solid use cases for AI in real estate

Here are 70 companies already ready to help you out.

All of this is very exciting but how do you ensure you remain on the right side of this change, and don’t lose your job to an AI?

Well first off you have to remember that Picasso, of course, was right. Computers ARE useless - they can only give us answers.

We are here to put the questions.

This is Gary Kasparov, the great chess master, and also the first world champion to lose to a computer, in his case IBM’s Deep Blue in 1996.

Since that day it was clear that humans would no longer beat computers at chess.

So Gary started a new form of chess, called Advanced Chess. In this a human, sometimes a pair, plus a computer would take on another computer plus a human or a pair.

What transpired was remarkable. First off a human + a computer always beat the computer on its own but also the team who managed the relationship between human and computer, regardless of strength, mostly came out on top.

As he says “weak human + machine + better process was superior to a strong computer alone and, more remarkable, superior to a strong human + machine + inferior process.”

Understanding how to best leverage the different skill-sets of humans and machines is the killer skill.

This was Steve Jobs at the launch of the iPad 2 in March 2011, and shows his famous marriage of ‘technology and the liberal arts’.

“It is in Apple's DNA that technology alone is not enough—it's technology married with liberal arts, married with the humanities, that yields us the results that make our heart sing.”

So, we have four steps to ensure your job is not taken away by an AI

Step 1 - DO NOT stay in a job that is ‘structured, repeatable or predictable’

This is from the Jack Lemmon film, The Apartment in 1960, and in those days offices were essentially Excel - every desk was like a cell in a spreadsheet where the person working at it was doing something ‘structured, repeatable or predictable.

If your job is anything like that, start planning your exit.

If your office is even anything like this definitely leave your company asap.

Step 2 - You need to learn how to think about AI

You need to Understand what AI is good for, and what it is not good for. Understanding where it can be applied is absolutely foundational and will avoid wasting a lot of time.

You need to look for those use cases - the 88% of core tasks from the RICS report, and the 49% McKinsey referenced.

Amongst those you need to consider which ones are commonplace and repeatable within your own business and/or applicable to many of your customers?

Then you need to think about what data you would need to feed your AI initiatives and whether or not you either have this data, or can acquire it in one way or another

You need to consider whether there are clear metrics by which you could judge success? Is success measurable?

And then finally, if you have all of the above, is that something that will create significant value?

If you get to the end of this with all the right answers then you have a product or service worth pursuing.

If you are looking to build AI solutions for customers then your product or service must abide by these rules.

Thirdly in a world of exponential technology you need to become exponentially human. You need to develop all your human skills - of imagination, empathy, design, social intelligence, problem solving, judgement etc - because that is your value.

Do not try and beat the machines at what they are good at - don’t bring a knife to a gunfight. Instead work on how, with your understanding of AI, you can use it to augment your own human skills.

As prediction becomes cheaper, the value of judgment will increase. All this technology is an input: we are responsible for deciding the output we desire. Those with great judgment, who can weigh up upsides and downsides of predictions will see their value rise.

AI will give us more options: how we exploit them will be down to human judgement.

And think ‘education, education, education’ - in a fast changing world education is a marathon, an ongoing process. You will never run out of new skills to learn.

And then finally you need to deeply consider how AI/Data and all technological change is going to change the world.

There will be winners and there will be losers. Oftentimes each of us will be both. We will lose some things but gain others.

Mostly it is tasks that will be lost rather than entire jobs but there certainly will be many jobs that do become redundant.

There will also be a lot of real estate that becomes redundant: as the work we do changes the nature of the office is going to change fundamentally. Where and how we live as well. And as for retail, well there is much change coming there.

Pay attention to those four rules and your job will not be overtaken by an AI. In fact you will most likely thrive in an AI world.

But either way remember that in business some things don’t change. 

The bear is coming but you do not have to be perfect, just better than your peers.

Previous
Previous

Rain, Steam and Speed V4: Why, and how, technology is upending the real estate value chain

Next
Next

Tweetstorm on OneMarket - Genius move or value chain disrupter? Both