Years after real estate firms first touted the potential of data science and machine learning to determine what tenants want, it’s not unusual now for
property companies to use artificial intelligence to analyze leases, communicate with tenants, value buildings, manage warehouse space and even woo investors. At a recent gathering of about 100 real estate professionals in New York, most attendees raised their hands when asked if their firms have used generative AI, an advanced version of the technology popularized by OpenAI’s ChatGPT that creates text, images, videos and other content after typing in prompts.
“We’re beginning to see a lot of use cases across the board,” Umar Riaz, who heads the Americas real estate, hospitality, and construction consulting practice at accounting giant EY, said at an Urban Land Institute New York event on AI and real estate. Within real estate, “there is no function within a company that is going to be unaffected by generative AI, all the way from marketing to operations” and research and development, he said.
Riaz added that the topic prompts questions whenever he travels to conferences and meetings across North America. While the real estate industry may not have adopted generative AI as “expansively” yet as the pharmaceutical or financial services industries, Riaz said it’s not “completely behind” and has a
“rich” user case.
“Everyone is uploading their leases into a generative AI tool to analyze the leases, that’s become very common,” he said, adding that a lot of real estate
companies are incorporating it for their chatbot functions to communicate with, say, apartment tenants. The technology is also beginning to be used for asset valuation, he said.
While generative AI has been touted for its growth across the globe that has led to major real estate owners including Blackstone to make major investments in data centers, Federal Reserve Chairman Jerome Powell said recently while there’s potential for the technology to increase productivity, it’s “probably not in the short run.”
There are also plenty of concerns about the technology. More than three quarters of senior IT leaders at companies said the tools bring “the potential for
security risks, and another 73% are concerned about biased outcomes,” according to a Harvard Business Review study last year.
For example, even though a generative AI chatbot giving incorrect steps for cooking a recipe may result in much lower “risk of harm,” the risk could be much
higher when it gives a field service worker instructions for repairing a piece of heavy machinery, the Harvard study said.
But even though generative AI’s lasting role in business remains uncertain, it’s important for landlords in the real estate industry to think of the potential
impact on tenant occupancy, Riaz told CoStar News. For instance, owners of buildings with heavy exposure to customer service call centers may need to see if they are vulnerable because of AI, he said. In another example, as companies use AI to better predict demand and reduce inventory, they may not need as much industrial warehouse space to stock inventory, he said.
“In real estate, we’re only as successful as our tenants and their businesses,” Riaz said. You have to look at “what industry your tenants are in. Is their
business up for disruption?”
AI Experimentation
New York real estate developer RXR has been using data science and machine learning through an in-house digital team it created several years ago to respond to tenant demands and study space use. Now the company is experimenting with generative AI to help in areas such as investor due diligence
questionnaires, which help interested parties with their evaluation prior to an investment or partnership, RXR Chief Financial Officer Michael O’Leary said.
“I grew up kind of in our capital raising and fund management world, but the company and I spent way too many late nights filling out these diligence
questionnaires for investors,” O’Leary said during a panel discussion at the ULI event. “We spent a lot of time responding to these questionnaires, crafting
answers, trying to sound professional and making sure our company comes out in the right light. Years ago, I tried to automate that a little bit. … I thought that was pretty savvy.”
But what generative AI has been able to do is what O’Leary described to CoStar News as “pretty wild” and “definitely the next level.” Generative AI is “going to help speed up our process and make people more efficient and allow them to focus on where we really need them to focus on,” he said.
While RXR primarily uses generative AI to automating mundane, time- consuming tasks, the developer also is studying how to incorporate the technology with its portfolio-wide property data to create its own version of a “chat asset manager” that users can ask questions of and pull data from, O’Leary said.
‘All About Efficiency’
Generative AI “is taking massive loads of information and understanding and producing a result,” Ajoy Bhattacharya, a Microsoft director and technology
strategist specializing in the architecture, engineering and construction sectors, said in an interview. Today, “it’s all about efficiency … [and] how do you make your job faster?”
For instance, generative AI can help summarize legal documents that are a big part of real estate, he said, adding it’s also being used to summarize emails and reports and make suggestions. A common industry use is what he described as “white book automation” to help, say, a developer look at different attributes of land and its marketability before deciding whether to invest in a project. “It’s only humanly possible to talk to a few people every day and retain even 10% of that,” he said, adding that generative AI “is really taking all that information that you can get from people and using 100% of it.”
For Savills, ChatGPT is among an arsenal of generative AI tools that help the brokerage “effectively” respond to requests for proposals, Patrick McGrath,
president of the firm’s west region, said at the ULI event. “That is a huge time suck for us. We have to do it with our data and we have to articulate the firm and how we think the firm is differentiated and why you should work with us. … That’s a great use case.”
At architecture firm Kohn Pedersen Fox Associates, the use of generative AI has become a daily affair in areas including wind simulation analysis, renderings and image making, which traditionally would have required hours of searching on Google, Joe Brennan, senior computational designer at KPF, said at the ULI event. The architecture firm’s goal this year is to build its own “custom generative AI tools,” he said.
“Over the last few months, we’ve seen it become really, really prevalent throughout” the firm, he said. For instance, project renderings traditionally are time consuming, especially with details such as adding in people and trees, and getting the lighting and reflections perfect, Brennan said.
“We’re not asking a computer to generate the design based off of that,” Brennan said in an interview. “Humans are really good at recognizing beauty and seeing beauty, and computers are really good at iterating and calculating things. [Things like] counting or testing or analyzing, we want computers to do that
stuff. And things that are more high-level design tasks or more creative tasks, we leave that up to the designers. … [AI] lets architects focus more just on the
design process.”
Article courtesy of CoStar News.