Over the last five years, I have been fortunate enough to publish several articles on Artificial Intelligence (AI), typically with a heretical tone. No one can deny that AI is powerful and that AI when applied deeply to verticals and scenarios will be a major factor in the transformation of business. On the other hand, the usual hyperbole that accompanies any trend in technology should not be taken at face-value; kicking the tires and assessing previous promises and bets is essential. To some, any suggestion that AI is anything but liberatory is heresy. But the track record suggests that, indeed, it is simply a statement of truth.
For the sake of simplicity, I will keep my comments restricted to residential real estate in the US. While there are many elements of this market that are extrapolatable to other markets, I will stick with what I know best. Also, for the sake of simplicity, I will take a complex supply chain and process and reduce it to five stages/processes:
1. Search
2. Valuation
3. Finance
4. Transaction
5. Post-purchase
For each category, I will give a “grade” (on the US scale, from A to F) and explain why I came to that conclusion.
Search: B
No doubt technology has enhanced the house search process- from the idea that most people initiate search from a mobile device to the fact that “Zillow” has become a household name. AI has advanced search and recommendation capabilities and has allowed companies to update the US footprint at breakneck speeds. Why this category is still a B and not an A is that much of the search process still uses human (agent) intercession and still requires local knowledge and wisdom provided by the agent. This is not an aspersion on agents, in fact it is the opposite, but AI here is still beat to the punch by human ingenuity.
Valuation: A-
House valuations, especially at scale, is a multivariate and complex problem. Here, AI is the perfect solution, offering algorithmic and computational nuances that consider hundreds of variables and account for the dynamism in these variables- in the forms of AVMs. In addition, the scale required to run the entire US footprint in hours (not weeks) requires an AI-basis as well. Valuation gets the highest grade and here, AI is a star.
Finance: D-
Housing finance has not structurally changed in decades. To the extent that large financial institutions have themselves undergone a sizeable digital transformation, technology certainly impinged on finance, but the basics are the same, untouched by AI. There are a set of companies, some established and some with success on the horizon, which look to create inventive models for housing finance, but they have not yet scaled. Put simply, AI has not delivered much to the finance process.
Transaction: D+
The transaction process is complex, often opaque to the consumer. While AI-led companies have certainly improved parts of the process, the entire beast is still serpentine and remains unenhanced by AI.
Post-purchase: F
Post-purchase, mortgage servicing remains imperfect, with scatter-shooting being the order of the day. Offers to refinance, leverage equity, and the like have improved but rarely does AI come into play, though it could help a great deal in targeting, data matching, and prediction. Finally, firms that work on maintenance and upkeep are still primitive in their technology approach.
This is my view on the industry at a point of time. It is a snapshot. I do not purport to have a crystal ball, but it is worth noting that there are far too many claims of AI in the industry and far too few real examples of customer value. Rhetorical must be replaced by reality if AI is to make a real dent in some or all the five categories discussed.
This column does not necessarily reflect the opinion of HousingWire’s editorial department and its owners.
To contact the editor responsible for this piece: zeb@hwmedia.com
