ULI Special Report: Getting Serious About Emerging Technologies
Artificial intelligence can be useful, but tailoring it to the industry's needs is key, said panelists during day one of the event.

ULI’s Spring Meeting is being held at the New York Hilton Midtown. Image by Fotios Tsarouhis
The evaluation of artificial intelligence and other advanced tech programs, as well as their role in multifamily, was a topic of discussion during the opening day of the Urban Land Institute’s spring meeting in New York City. Panelists shared insights into how AI can be best utilized when it is tailored to real estate needs.
“Real estate, as an industry, is certainly an extreme laggard relative to many other industries in adopting more systematic approaches to data science and machine learning,” said Drew Conway, head of data science, private investments, at Two Sigma Investments. “But to me, the question is ‘Why?’”
READ ALSO: ULI Spring Meeting Highlights
While these programs have been more intricately integrated within other sectors, there has been resistance in real estate, said Conway. Part of it may be the personal nature of the industry, in which tech may be seen as second to human connection. “It’s a highly personal, in-person business. It’s a highly tactile business, one where things are changing very rapidly on the ground.”
The incorporation of data and machine learning can be a boon to multifamily and real estate more broadly, but only if done properly.
“Most of the things that I’ve seen in real estate are the incremental improvements over current processes that are efficiency-based and do things a little bit better,” said Josh Panknin, director of real estate artificial intelligence research & innovation at Columbia University.
Pankin stressed that it is not enough to hire a data scientist unfamiliar with the nuances of real estate and expect that person to come up with acceptable products or solutions. “It doesn’t work that way if the machine learning engineer or programmer has zero experience.”
Those who are serious about utilizing AI must understand that the most efficient approach is identify which problems in real estate are best suited to be solved with these technologies, said Panknin. “AI is not magic, it’s mathematics. Everything has to be quantified.”
Data scientists and machine learning engineers look at things in a very structural way, “but they often don’t look at them in a way that represents the nuances and idiosyncrasies in the heterogeneity of real estate,” said Panknin. “There has to be a computational way, a quantitative way, that real estate problems are structured in order for AI and machine learning to be effective.”
“Real estate has, in my opinion, taken a very lazy approach,” said Conway. “It’s real estate people’s responsibility to learn how to use these tools, learn how to structure them, learn how to put them together so that engineers that don’t have a background in this field can build models, build tools that fit what we’re trying to do.”
Building vs. buying
What aspects of a company’s AI and machine learning infrastructure should be built internally versus how much can be purchased is emerging as an important issue. Contemporary tools that have become “modern AI,” such as language models and generative AI, do not need to be built internally, said Conway. Attaching a research development process to a specific business outcome, however, is a linkage that needs to be done internally with intimate knowledge of how the product functions. “Ultimately, a decision to deploy capital is going to be made based on the results of that work.”
“If it’s something that’s industry agnostic, (such as) HR software, accounting software, it doesn’t make any difference,” said Panknin. “If it is core to your business, how you provide value, efficiency-based insights … if it’s something that is key to how you do business and how you can gain a competitive advantage,” then it is important to build the technology in-house.
Designing role
AI and new technologies are creating new possibilities in design, a burst of modernity for the staid multifamily sector. “Our industry really hasn’t faced massive disruption,” said Gensler Co-CEO Jordan Goldstein on a later panel. “When I got into this profession in the mid-90s, it was right at the time that CAD (computer-aided design) and digital design was really coming into play, so the usage of 3-D modeling rendering animation was starting to become commonplace.”
Now AI and related technologies are contributing to the changing landscape of property design. What the new tech will be used for, however, remains unclear. “Here we are with a full set of new tools,” said Goldstein. “AI, if used properly … can accelerate the design process and really challenge us creatively.” Technologies like spatial computing, too, will “change how we are able to think about built environment long before it’s built.”
“With regards to AI, you know, it is interesting, because we’ve been recognizing that there’s going to be big change coming with regards to this,” said Goldstein. So rather than being reactive to it, how do we be proactive and really start to test out ways to impact the design process?” One innovation so far is AI programs that can react to, and offer improvements on, prompts for multifamily properties. “It also allows us to really think differently about data, data as a currency.”