AI in Proptech: Reality and hype
There is a defining concept in every era of business and technology. Innovations tend to be clustered around these concepts; as they emerge into the market, they create a flywheel that further elevates the idea. As investors take note, they pour money and time into it and adjacent areas and the pie continues to get bigger. The defining concept of this decade is AI and the large pie has uneven consistency and taste.
In less metaphorical language: There is a lot of hype around AI and some reality as well.
Proptech is no different here than any other area of the business world. As the notion of a “digital ecosystem” in the real estate/mortgage/financial world takes root, and as technology investments are seen as the silver bullet to transform what is thought of as a “traditional” – even stodgy – industry into a dynamic and agile digital marketplace, claims about AI abound.
This article is not about exposing particular companies’ claims or anointing winners in the race, but, instead, is about the approach taken. The approach and the processes that follow are the best determinant of future success or failure, are the most clearly indicative of reality versus hype.
Take the area of residential real estate- the world’s largest asset class. In the US alone, the aggregate value of residential real estate is $35 trillion. If you take the forward 12 months as an example, over 6 million houses will sell and there will likely be over 15 million monetizable transactions of some sort or another. Put simply- residential real estate is big business.
Still, some basic questions continue to vex the industry. Why is House X valued at a certain amount and House Y valued at a different amount? Why do some types of houses gain in value more than others? Why do different parameters have different weightings in disparate geographies? These are natural questions and ones that can only be answered with data modeling, AI, and machine learning. But the approach taken has to be one of asking the relevant questions first and letting the data and AI follow.
The residential real estate market also has a lot of “outliers” and “edge-cases” that have to be accounted for. A good AI scientist has to understand this as an input in the system and not just an output. The best amongst them will create algorithms that mimic other similar systems – say the systems of genetic mutation- and then modify them with the practical aspects of real estate in mind. Again, this is a question of approach.
AI thus cannot be an afterthought, an after-market bolt-on. That makes for great rhetoric and poor reality. AI has to be inherent.
The world of real estate, mortgages, and property, in general, is going through changes- at a scale and speed heretofore unseen. A variety of tools, processes, and approaches are available to the industry.
Scatter-shooting and after-the-fact declarations don’t do any good. I’d encourage all organizations in this vast and wonderful ecosystem to ask the right questions first and then determine the answers with the innovative new tools and processes available. Separating hype from reality isn’t easy but it is necessary.