Leases  ·  Sales

How GenAI Threatens the Use of Office Space in the U.S.

Generative artificial intelligence could both damage and bolster demand for conventional offices

reprints


As one of the first films to predict the birth of artificial intelligence, the early 1990s blockbuster “Terminator 2: Judgment Day” opens into the dystopian future of 2029 Los Angeles, with sentient, super-intelligent machines waging a violent war against what’s left of humanity. 

For some office owners, director James Cameron might have been anticipating the office market in 2024.

SEE ALSO: L+M Plans 328-Unit Affordable Housing Development at 1225 Gerard Avenue in the Bronx

While not quite as dangerous to the species as Hollywood once predicted, the rise of generative artificial intelligence (GenAI) — computerized machine learning that simulates human intelligence, language processing, problem solving and decision making — has changed the face of work and is likely to impact foundational jobs within white-collar professions.

Assistants. Administrators. Lawyers. Accountants. Even, dare we say, reporters. They could all be at risk. An expensive salary today now appears expendable tomorrow when a supercomputer can do the same job at a fraction of the cost. And that will mean a change far more radical to the nature of work than anything from COVID-19.

A 2023 research paper from Goldman Sachs economists predicted that roughly two-thirds of jobs in the U.S. and Europe are at risk of some degree of AI automation, with GenAI technology threatening to replace one-fourth of current occupations — placing 300 million jobs at risk of automation between the two continents. 

“In terms of the exact order of magnitude, it’s hard to know,” cautioned Sarah Liu, partner in the real estate technology arm of Fifth Wall, a proptech venture capital firm. “Overall, you can trust that AI will result in fewer people required in a number of white-collar jobs, whether that’s sales, marketing, writing, lawyers or accountants.” 

In New York City, the heart of the nation’s economy, the threat of GenAI is already turning heads. A March 2024 report from McKinsey concluded that by 2030 as many as 1.1 million occupational shifts will be needed across the five boroughs, with 380,000 of those job changes a direct result of AI automation. 

Harry J. Holzer, professor of public policy at Georgetown University, believes that young people entering the workforce, particularly college-educated workers, are among the most vulnerable to the promises and perils of GenAI, particularly the language learning models like ChatGPT, DALL-E, and LaMDA that can simulate speech, writing and human thought. 

“You can imagine a program that writes very well could be a substitute for journalists or paralegals or research assistants,” explained Holzer. “A lot of technology that produces this can hit the low-end college grads harder than other people. They are involved in jobs that do a fair amount of standard writing and low-level data manipulation that AI does better than them. 

“It means that some chunk of early entry jobs might no longer exist,” he added. 

Scott Rechler, CEO and chairman of RXR, a New York City real estate development firm, said that the rise of AI means some back office administrative functionality “is going to be less people intensive” because of the new technology. He also said, though, that AI will enable employers to get rid of “soul-crushing tasks” that people in white-collar fields don’t find energizing or engaging. 

“It will enable those people to do more value-added type of work,” said Rechler. “Then the counter argument [to the occupational shifts] is that it’s more of a transition to retrain people, so then you’re putting people back into the workforce with other responsibilities and value-added tasks that’s more enriching to them and more productive to the economy.” 

But the unspoken issue at the heart of the entire AI transformation in America is what precisely  will happen to where people work, and what will become of how people work, now that we’ve made a world where a supercomputer can replace an employee? In other words: What is the role of traditional — even highly amenitized, newly constructed — office space in this soon-to-arrive future when machines steal the jobs of humans?  

“For office space, and for most real estate, you can go back to fundamentals of supply and demand: Whenever you have anything that will reduce demand, that’s bad for real estate,” said Chad Carpenter, founder of Reven Capital, a real estate investment firm. “When demand decreases, that puts downward pressure on rents and values.” 

Even Rechler, a longtime bull on the value of office even through the upheavals of the COVID-19 work-from-home paradigm shift, admitted some concern around the threat GenAI poses to the bottom line for traditional office space sponsors. 

“I’m less worried about hybrid work than I am about AI,” said Rechler. “Because hybrid work has played itself out. People need to be in the office, they want to be in the office, they recognize the value of that. But the point of AI is there’s an element that won’t need to be in the office, or even exist.”

Despite opening Reven Office REIT, a publicly traded real estate investment trust devoted to investing debt and equity into distressed office assets, Carpenter is among the bears when it comes to office usage in a future economy littered by GenAI bots. 

He compared the eventual decline in some office use to what is now happening across retail and fast food stores — where one can go into a McDonalds to begin self-ordering — and almost all banking footprints  — where online banking has reduced the need for 5,000-square-foot physical spaces with huge vaults, dozens of tellers and numerous desks. 

“As AI technology gets better, less people will be needed, and that will have a big impact on the Class A and B office space for legal, financial, insurance and engineering firms,” said Carpenter. “But I think it’s the back office firms — call centers, settlements, clearance, record maintenance, regulatory compliance, accounting, IT services, stuff like that — they’ll really get hit with less people.”  

It’s not like this should come as a surprise. As far back as November 2019, months before the pandemic upended work in America and years before GenAI became a household name, the Brookings Institute published a research paper that concluded AI could affect work “in virtually every occupation group,” with highly educated workers particularly threatened by the new technology. 

“Our analysis shows that workers with graduate or professional degrees will be almost four times as exposed to AI as workers with just a high school degree,” noted the report. “Holders of bachelor’s degrees will be the most exposed by education level, more than five times as exposed to AI than workers with just a high school degree.” 

For an office sector that is presently experiencing an availability rate of roughly 20 percent (the highest on record), the widespread adoption of GenAI comes after millions of square feet have already been called into question through nearly half a decade of ingrained work-from-home practices by employees and employers. 

Carpenter calculated that as lease rollovers play out in the office space, firms could reduce their office footprint by another 20 percent over the next five years, bringing the office availability rate to as high as 40 percent.  

“And we haven’t even gotten to AI yet,” he said. “If Goldman Sachs (GS) is predicting 18 percent of the workforce won’t be needed and 10 percent of that is for office, now we’re at a 50 percent vacancy rate. Will it happen? I don’t know. Could it happen? Absolutely.”

‘I’ll be back’

Just as Arnold Schwarzenegger was a bad AI cyborg in the first “Terminator,” he was a good one in “Terminator 2.”

For all the talk of the total pie of white-collars jobs shrinking due to GenAI, many within the real estate and proptech spaces believe there will also be counterbalances, and that not only will thriving AI companies need to employ people to manage the fleet of bots, but companies that haven’t yet adopted AI will hire more people because of the new technology. 

Aaron Block, co-founder and managing partner of MetaProp, a venture capital firm focused on proptech, conceded that GenAI’s advent will require fewer humans for some physical work and thus reduce floor plates; however, the technology is a double-edged sword for real estate and white-collar industry. 

“What doesn’t get talked about enough is the other side of the sword: the new businesses and new hires this efficiency will enable,” Block said. “It’s not one story, it’s several stories: It’s market dependent and it’s industry dependent.” 

Block compared the embrace of GenAI to the birth of the calculator and personal computer. Even equipped with calculators and personal computers, we’re still doing math in an office environment, but how we’re doing this math and what we’re using to do the math has evolved.   

“And we’ll see that evolve in other tasks beyond math,” he said. “I don’t know how much of the reduction in space due to AI taking human jobs will be offset by the new jobs or businesses created to fill that space, but I have a feeling it will be on par or better for the new job growth creation in markets that are important, like New York City.”  

Block is not alone in seeing opportunity where others sense only disorder. 

Yao Morin, chief technology officer at JLL (JLL), believes that the demand for AI and the growth of AI-related firms will translate into a demand for office space across the U.S. 

Morin said that JLL research found that, at the end of 2023, 1.9 million square feet of office space was in demand nationally for AI-related firms seeking to grow their physical footprint, double the demand from those same firms in 2020. Some 42 percent of that demand was concentrated in the Bay Area, followed by large concentrations in Boston and New York.  

“There’s tremendous demand for office areas in these tech hubs, and we believe you will first see office market demand go up in the major tech hubs and then the secondary markets will follow,” said Morin. “And it will continue growing from a demand perspective because job postings will increase concentrations for AI talent.” 

Morin also brushed aside fears of GenAI automation supplanting jobs in the current workforce due to the fact that, in her view, GenAI is still very far from replacement-level quality, even extrapolated out over three to five years. She compared it to the last-mile problems that self-driving cars are experiencing, and the persistent need for a driver to remain behind the wheel. 

“AI is facing similar problems,” she said. “You can augment, you can facilitate a lot of jobs. However, it still requires human endeavor for validation and supervision and quality control. So I don’t see it having an impact on white-collar jobs.” 

Others heavily involved in day-to-day office space research and sales are similarly skeptical about the eventual negative influence GenAI will have on white-collar jobs and office space. 

Michael Lirtzman is head of U.S. office agency leasing for Colliers (CIGI). Lirtzman calls the concern around GenAI replacing physical jobs “a fear, and probably an unjustified and unrealistic fear at this point,” largely due to the human touch required for so much white-collar work, particularly in commercial real estate. 

“In our industry, the majority of transactions that get done — leasing, acquisitions, dispositions, the consultants that work on it, the lawyers, the accountants — there are personal skills and personal relationships that go into it,” he said. “Those things can’t be replaced by technology.”

Even if millions of jobs are replaced, primarily in white-collar office work, that is merely a reflection of modern economic history, according to labor historians. Georgetown’s Holzer notes the number of jobs that exist in 2024 that didn’t exist in 1970, or even appear on the radar, and how many of those are a result of technological innovation.

“AI will do that as well, creating all sorts of new categories we aren’t even aware of right now,” he said. “It will improve health care diagnosis, financial diagnosis, legal diagnosis, so there will be big positives. 

“When productivity goes up, and costs come down, it benefits the entire population — the question is how many people will face this displacement and how will they adjust,” Holzer noted.   

This displacement, however, is the crux of the equation. Even the most optimistic, if not rosy, assessments of AI always come with a job-killing caveat.  

“For the most part, we don’t see AI replacing people entirely, we see AI enhancing people’s productivity,” said Fifth Wall’s Liu. “Rather than 10 people needed to do that job before, you only need two people because you quintupled your efficiency. 

“So, overall, there will be plenty of white-collar jobs and need for office space, but there will be noticeable footprint shifts for certain sectors where it’s easier for AI to replace 50 to 60 percent of the job efficiency,” she added.   

A new future 

Regardless of how many jobs are automated out of the white-collar workforce by GenAI, what’s indisputable is that the technology will change the face of the U.S. economy. Experts in both proptech and commercial real estate believe that GenAI is already changing office buildings themselves — and largely for the better. 

Fifth Wall’s Liu listed a litany of ways GenAI will change landlord and tenant office space usage in the 21st century: improved access controls and tenant engagement controls; computer vision creating new ways to analyze how tenants use space, leading to more effective layouts; GenAI generating floor plans and providing sharper 3D tours to prospective tenants for spaces that haven’t yet been developed or renovated. 

“It will make office buildings just continuously smarter,” she said. “And I think the sales side in office will actually benefit quite a lot from GenAI in being able to help tenants envision how they use that space and use the space more productively.” 

In New York, RXR’s Rechler has already been implementing advanced AI technology into the dozens of office buildings his company owns across the metropolitan area. Rechler told Commercial Observer that he began rolling out AI as far back as 2020 to improve and monitor COVID-era health and wellness compliance and social distancing in his office towers. 

“It was a huge group of programs but they weren’t large language models,” Rechler explained, noting that his buildings today can codify sustainability and track individual unit usage to make tenants more energy efficient. “We’ve been doing that for a while here, and [the AI] will just get better and better at being able to consume a significant amount of data about what’s happening in your buildings and with customers to create a better level of service, better efficiency, and an optimization of energy.”  

These types of GenAI algorithms that predict power usage and other building utility designs have been around for many years, according to MetaProp’s Block, but it is only in the mid-2020s that they are being taken up by owners conscious of efficiencies and keen on reducing capital costs. 

“While none of this stuff is new, it affects not only operators of buildings, but also the boardroom — how much you budget for capital improvements, repairs and maintenance,” said Block. “So we’ll see more and more of that application of hardware and software that allows you greater transparency to get ahead of things that haven’t yet happened, but might in the future, and nip them in the bud.”   

Forget building footprints or faulty air-conditioning systems, even the work itself within CRE firms is changing under the guidance of GenAI technology. 

L.D. Salmanson, co-founder and CEO of Cherre, a leading real estate data insight firm, told CO that AI has already modified the white-collar functions of the CRE sector, and that probabilistic models have replaced outdated thinking built around hunches and relationships when it comes to investment decisions.  

“I don’t get to make assumptions anymore; models are driving the assumptions. I don’t get to say, ‘I like a market,’ the models are telling us which market is more attractive and less attractive,” said Salmanson. “Instead of guessing, I have a model that fills in the most probabilistic answer. That’s what’s happening today  — all things are becoming probabilistic and model-driven.” 

The largest operators in CRE capital markets investment — Starwood, Brookfield, Blackstone and the like — are already well ahead of the curve in figuring out where to invest based on GenAI analyzing decades of firm data and past decision-making processes, according to Salmanson. 

“They have already connected all their past data, they can report on all their assets worldwide, and they can identify past trends of where they did well versus their peers and why,” he explained. “Now they can answer, ‘I’m really interested in that market. How might that look?’ ‘What if I did XYZ and rates went down. What does that strategy look like?’ They’re able to do that today.”

In many ways, in this new survival of the fittest abetted by supercomputer technology, whichever CRE firms adapt to GenAI first will separate themselves from the pack. 

“Large clients are already generating alpha by identifying markets and submarkets and strategies their peers aren’t able to deliver,” explained Salmanson. “It’s now about, ‘Given all the info I have, my model is predicting growth where others are not, so I’ll invest there today when others are not. I’ll generate alpha where others are not.’ That’s the most interesting part — the moneyball experience.”   

Long game

Dreams of moneyball notwithstanding, commercial real estate is in the midst of a major structural shift in how office space is used and which employees will remain part of the 21st century workforce. Particularly because so many firms, across all sectors and industries, are adopting GenAI into their business models, many sense that it’s only a matter of time before widespread disruption takes place and the value hits to office investment follow. 

Peter Weiss, co-founder and managing partner at Alpaca Real Estate, a CRE private equity firm, said that his company’s aversion to office comes from its own enthusiastic use of GenAI technology and what they extrapolate from it in the ensuing decades. 

“If we’re doing these things, and if you forecast for the next 10 to 20 to 30 years, our guess is other firms will use this technology to their advantage, too, and there will be efficiencies at scale, and that will likely lead to using office space in a different way,” he explained. 

To this end, Weiss said Alpaca Real Estate is not investing in office buildings, largely due to the fact that most office leases are 10 years in duration. Seeing as we’re only four years into the post-COVID shift in work patterns, at least 60 percent of firms leasing office space haven’t yet decided how they’ll use their space going forward amid hybrid work and GenAI adoption. 

“We don’t think you can prove demand in office today,” he told CO. “We think we are so early in this change in how office is being used, changes in work from home, there are a lot of changes moving through the system. … It’s just not an area that we’re excited about from a return perspective.” 

Even an aggressive office investor like Reven Capital’s Carpenter is hunkering down and preparing himself and his investors for a decades-long play to reap the rewards of distressed office values. Carpenter said he is cautioning clients to underwrite longer downtimes to lease-up and to forecast lower rental rates in the office space. 

“I think the space could go vacant for a very long time,” he said. “I’ve been the conservative guy, so let’s play this out for 10 years, maybe 20 years, not just two years because it could take a lot longer to lease these buildings than people think.” 

The silver lining in all this disruption, according to Carpenter, is that tenants will still seek space amid the value declines, and the fewer choices available among surviving footprints will create an increasingly competitive landscape.

“Office is going to be a pure commodity product, and you’ll have to drop your rents and offer big concessions and TI packages to steal tenants,” he said. “But you can do it, especially if your basis is a fraction of your neighbor’s.”

As for the workers, well, until the GenAI models can argue a case in court, or lead a tenant negotiation, or woo investors to commit millions in capital, the use of human beings is likely to continue to drive CRE work and the office market as a whole — at least until Judgment Day arrives. 

“The reality is models need constant training, and beyond that, as information evolves they need to evolve as well,” said Ermengarde Jabir, senior economist at Moody’s. “In theory, could everyone one day be replaced by AI, and the only people left working are the engineers that build AI? In a dystopian world, that’s possible. But will the AI show up at the courthouse?

“The reality is that won’t be the case,” she said.

Brian Pascus can be reached at bpascus@commercialobserver.com