Has Generative AI Had a Practical Impact on CRE Yet?

The hype cycle for generative AI—artificial intelligence that scoops up vast amounts of content, whether text, image, computer code, or other—has been immense. Rarely, if ever, has a relatively new technology gained the popular fascination, claims of ubiquitous applications, and growing cadre of self-proclaimed experts on how to use the products, even though the software has been available for a relatively short time.

Many CRE-focused software companies have flocked to the capabilities. “Generative AI has tremendous potential in the commercial real estate industry, an industry that has generally been slow to adopt and embrace technology,” FTI Consulting senior managing director Josh Herrenkohl tells GlobeSt.com.

Potential, though, has little meaning for companies when the question is where such software has had a practical impact on CRE already.

Matt Rossman, director of marketing at Bryan Industrial Properties—which owns, brokers, and manages almost 2 million square feet of industrial property in Orange County, California—tells GlobeSt.com, “Generative AI has been an incredible resource for us as it allows us to focus our time on productive work and run the portfolio with a very compact team.”

The company uses ChatGPT, DALL-E, “and a few other AI platforms in several ways.” They include creating letters and correspondence, first drafts of marketing materials, lease abstraction, social media image creation, email campaigns, and even writing some custom software to review their portfolio health and status.

Going further can be difficult because of the current state of generative AI and some choices the vendors make. “We at Future Today have been looking at AI for years now and help our clients understand the implications,” says Mark Bryan, a senior foresight manager at the Future Today Institute. “But most of the time, they fall into the hype cycle. They are continuously reacting versus being proactive,” like seeing a competitor use something and reactively adopting it. “They take it for granted without realizing how AI is operating and functioning.” That can be a problem when the AI vendors keep the software models and the training data secret, making strategic decisions difficult to formulate.

“From a generative ai perspective in real estate, we’re barely scratching the surface,” says Ram Srinivasan, managing director of the global consulting group within JLL, which itself has started using ChatGPT. Companies are investing huge amounts of resources. “There’s a lot of buzz around it and a lot of investment dollars chasing it, and a lot of executive attention on it.”

But it’s all new and the full fruit is likely sometime in the future. Part of making effective use of them today is understanding the limitations. “The power of these applications is that you can really move the productivity needle significantly,” Srinivasan says. “Where people get tripped up, these models are not research tools.” There is a phenomenon called hallucination in which generative AI, in response to a prompt, will string together parts for an output that is utter imagination.

Instead, currently the better applications are as “a content partner” to create new ideas, scenarios, and outputs, as Srinivasan explains. “Things like can you create scenarios for occupancy planning/? Can you support this new vision statement? Design options for floor X? These are areas where generative AI’s power holds supreme and we’re seeing applications already.”

“Specifically in Gen AI, there’s been a bunch in multifamily and chat bots for prospective resident capture,” says Deloitte managing director and real estate solutions leader John D’Angelo to GlobeSt.com. “The real attraction has been interaction with prospective residents and viewings. We’re seeing real deployment,” particularly in larger multifamily companies.

Another large Deloitte client manages 30,000 active leases and is using generative AI to analyze contracts. Some of the data has to be abstracted from the text, whether numbers and conditions or more complex aspects like restrictive uses in retail for a given tenant or property or location. Ordinarily, someone would have to pull up the contracts and manually read them. Software can cut hours out of the process.

“There’s such a pot of gold at the end of that rainbow of not just changing how you get access to data but how you work with it,” D’Angelo says. “The things that seem particularly promising are site selection, being able to use a tool with large available datasets and insert my requirements. That seems like it’s got legs. The specific lease generation stuff, having that be a product, that seems promising too.”

But much of these more expansive practical uses is among larger companies rather than smaller because of the clear resource gaps. They may not have the in-house knowledge, money, time, people, or attention to find increased uses of the technology. “It’s harder for smaller companies to do the same investment,” D’Angelo points out. Which means most small companies won’t get some of the best available benefits of the technology.

There are also legal problems on the rapidly approaching horizon. The New York Times has a copyright lawsuit targeting OpenAI. If found liable, OpenAI could find itself facing massive legal fines and a need to license material at significant cost, also raising the question of where other AI companies stand and where they get their data. As OpenAI said in a filing with the U.K. Parliament, “Because copyright today covers virtually every sort of human expression–including blog posts, photographs, forum posts, scraps of software code, and government documents–it would be impossible to train today’s leading AI models without using copyrighted materials. Limiting training data to public domain books and drawings created more than a century ago might yield an interesting experiment, but would not provide AI systems that meet the needs of today’s citizens.”

Bruce Stachenfeld, chairman of Adler & Stachenfeld, thinks that people in real estate need to change their assumptions to make AI truly valuable. “It’s not a new business, “It’s a new way of doing business that I think will be very useful to people who know how to use it and dangerous to those who don’t.”