Finding an AI equilibrium

Like many industries, private real estate is seeing a wave of artificial intelligence-driven innovation, and few can refute the disruptive potential of the technology.

AI-based programs are being used to identify acquisition targets, underwrite deals and manage assets. Complex processes such as entitling land and financing affordable housing are being expedited by algorithms. Some upstart firms are even basing entire strategies around the use of machine learning to help institutional capital reach previously unattainable properties.

Clelia Warburg Peters, managing partner of Era Ventures, a venture capital firm focused on innovations in the built environment, says such firms should not be viewed as a niche subset of managers, but rather as a model for the future of the field.

“The internet is such a fundamental part of the way we operate in the world today; it would be weird to talk about a ‘web-based’ business. AI is going to be similar a decade from now,” Warburg Peters says. “It will be weird to talk about an AI business because AI will be a – if not the – fundamental component of how businesses deal with data and manual processes. It will serve as the first line-of-defense interface for most touch points between companies and customers.”

The next generation

Private real estate firms have used basic algorithmic tools for years to expedite processes. But the latest iteration of this technology is known as generative AI, as it can create unique content. This leap in capabilities has been driven by the development of large language models, or programs that can process human speech at a sophisticated level. The byproducts of this are systems that can gather information with the speed of a computer and apply it to a task with the precision of a person.

“The internet is such a fundamental part of the way we operate… it would be weird to talk about a ‘web-based’ business. AI is going to be similar a decade from now”

Clelia Warburg Peters
Era Venture

“Within the generative AI space, there is really a broad range of opportunity for people either to be more efficient in their work, or in some cases, really streamline the amount of man or woman hours it takes to get things done,” says Jake Fingert, managing partner at the proptech investment firm Camber Creek.

Typically, these algorithms are programmed to compile raw data from disparate documents and datasets and synthesize them into something coherent and useable. Ethan Chernofsky, senior vice-president of marketing at the foot traffic analytics firm Placer.ai, says AI can quantify market observations that have until now been difficult to measure.

“So much of the real estate industry is based on hunches, but the interesting thing about that is a hunch is another way of saying a reaction to data,” Chernofsky says. “A gut feeling does not come from nowhere. It is a reaction to input signals. A lot of what data is doing is allowing investors and other players in the physical world to put numbers and metrics behind those observations.”

Embrace and educate

Yet, the potential cost savings that AI can bring comes with significant risks and uncertainties, particularly around AI-based decision-making.

Such fears were crystalized in November 2021 when Zillow, the real estate listings giant, shuttered its algorithmic home-purchasing system after it generated more than $1 billion of losses in less than four years. The program’s faltering has widely been attributed to its inability to properly price assets.

Christy Fields, managing principal and head of real estate portfolio solutions at the consultancy Meketa Group, says the ordeal has not soured institutional investors on AI, but it has contributed to a hesitancy about diving headfirst into AI-centric strategies.

“People are in a little bit of a wait-and-see mode around AI. There is some skepticism around its ability to predict pricing, among other things,” Fields says. “Working this data into site acquisitions or property acquisition or site acquisition efforts is pretty interesting, but we do not have any realized track record around it. We cannot say: ‘You did this and you’re producing better risk-adjusted returns.’”

Even firms that have placed AI at the center of their investment strategies are not banking on it replacing human expertise entirely. One such group is Keyway, a New York-based manager that uses artificial intelligence to identify acquisitions.

The firm’s platform can identify properties that are available for sale as well as those that soon could be – based on factors such as maturing debt or owner credit defaults. It can also determine whether a property is ripe for a value-add strategy by looking at occupancy, nearby market rents and when major renovations last occurred. It can even compose offers.

Matias Recchia, co-founder and chief executive of Keyway, says the firm’s technology enables institutional investors and their managers to deploy capital into smaller assets by reducing the time and cost of underwriting them. But, he stresses, all deals are reviewed by experienced real estate professionals and all decisions are made by humans.

“AI is using historical data for the most part, it is not predicting the future,” Recchia says. “When a six-sigma event or an eight-sigma event happens, you need humans there to do a sanity check, to understand the context of what’s going on.”

Also, because algorithms draw from historical inputs, they can be shaped by the biases of the past or even the present. In the context of approving tenants or loans, Recchia says this can cause programs to discriminate against certain groups based on the factors they have been taught to consider high risk.

“When it comes to background checking renters or approving loans for first-time homebuyers or platforms geared toward retail investors, there’s a lot of regulation, thoughtfulness and checks that need to be put around AI,” he says.

There is much information for investors, managers and service providers alike to absorb about AI, and many risks to consider. But Warburg Peters says industry participants should not feel compelled to draw conclusions about the technology today. Instead, she says, they should be aware of it and open to incorporating it when and where it makes sense.

“The reality is that for many people, the right choice right now is to educate themselves, not to be afraid, and to watch it and see how it has evolved,” Warburg Peters says. “It’s not like this is, for most people in the real estate world, a gold rush that they are going to miss out on if they don’t get involved. It is going to be a part of everything pretty rapidly.”