Your Dream House Might Be a Fantasy

AZillow listing in Detroit recently went viral— and not for the usual reasons. Rather than boasting an extravagant interior or other routine signs of gluttony, the house itself was quite ordinary: a warmly painted house with a beautiful dormer window and a cozy veranda. Or so it seemed. When netizens viewed the ad on Google Street View, it soon became apparent that the house's images had been carefully manipulated using artificial intelligence. The dormer window is gone, the façade is clean and freshly painted. (There is a dormer window, but it is pointed, that is, of a completely different shape.) Even the courtyard in the ad photo sounded creepy: too green, flat, like in the illustration. The actual home, as it exists in the physical world, is, to put it mildly, a fixer-upper top at best, with chipped paint, outdated aluminum-paneled windows, and a general appearance of wear and tear. Perhaps more shocking than the exterior facelift were the interior images on the listing, each glowing like AI images with a filter haze. The house itself was not AI-generated—that would be cheating, after all—but what was real and unreal was mixed in a slurry of ambiguity, truth and lies.

Over the past couple of years, a number of artificial intelligence services, such as virtualstagingai.app and video generator AutoReel, have infiltrated the real estate industry for the same reason they have infiltrated every other creative industry: to disrupt or eliminate the work of skilled workers. However, rather than eliminating photographers and stage managers in one fell swoop, as Uber did with taxi drivers, the introduction of AI is simply accelerating a long-term labor trend in the real estate industry, where real estate agents have long been motivated to eliminate outside contractors in the pursuit of increasing their profits.

While the use of artificial intelligence may seem particularly egregious in its laziness, real estate listings have gradually become increasingly artificial over the decades, starting with the spread of Photoshop from the desktops of professional photographers to ordinary people in the noughties. This shift caused a proliferation of images that used high dynamic range exposures, where every edge was too sharp and the sky was too blue. By the dawn of the aggregator era (Zillow, Redfin, etc.), blatant image manipulation had become the norm. Given the economic conditions, this was almost inevitable.

We take it for granted that we can log on to the Internet, and since founding Zillow in 2006, we instantly know the estimated value of any home. This development is arguably responsible for a number of social ills, from predatory flippers closing in on vulnerable homeowners to a simple loss of privacy. (It could be said that the first skilled real estate worker the tech industry cut was the home appraiser.) Before the era of aggregators, real estate listings were run by agencies and only accessible through the Multiple Listing Service, proprietary indexing agents that were once kept under lock and key. After agents agreed to share their data with aggregators to make their listings more competitive, the real estate site as we know it today was born – an endless stream of houses.

According to Zillow CEO Spencer Rascoff in his book Zillow Discussion (yes, really), it was intended to “democratize” the real estate industry, freeing it from a handful of pesky gatekeepers. In fact, it has turned what was already a highly competitive field into a more exploitative one in which every effort must be made to outdo others in the now transparent open market. This rat race, ironically for agents, devalues ​​many of their previously valued skills, as well as the skills of the contractors they hire. Soft skills such as writing and giving tours were among the first.

For example, when two houses, each sold by different agents, are equal in both price and amenities, the sale comes down to how each is presented in this information-rich space. This pressure forces agents to resort to spectacular or deceptive behavior to gain an advantage over their competitors. These days, they use everything from drone footage, 3D renderings and walkthroughs, digital staging, and (of course) image enhancement of varying degrees of quality. (I often doubt that a sunset conveniently located behind a certain estate in Naperville, Illinois, can truly rival that of Bali.)

These developments, and with them the expectation that homes should look so perfect as to test the boundaries of reality, have already paved the way for the use of artificial intelligence tools. Now AI is poised to wipe out one of the industry's last skilled workers: the real estate photographer. A good real estate photographer can charge hundreds or even thousands of dollars per shoot, depending on the size of the property, often not including installation costs. Photographers have so far managed to adapt to new trends by using wider lenses and apertures to provide the bright, diffused light needed to make digitally created furniture work best. But if an agent, homeowner, or renter had the ability to take photos on their phones and had AI enhance them in a way that was even remotely plausible—given the already lowered standards of digital real estate economics—it would be an extinction-level event for one of the last ways to make a living as a photographer. This will have a ripple effect on the rest of the industry, including more agent work, the end of a viable industry outside of luxury markets, and potentially legal implications as the lines between boosted and false advertising blur.

At the moment the technology is not quite there. In the end, we can still recognize the difference between an AI image and the “real” one. The characteristic yellow tint of AI-generated images, free generators that produce images with eerie brightness, and constant errors introduced by algorithms (like rearranged dormers in our Detroit example) are all problems that may hold AI back for now. But will they keep it under control forever?

What happens when we can no longer trust the images we see online? This is already a relevant question in relation to AI in general, but in relation to real estate in particular it has not been considered in detail. I think it's entirely reasonable, given the industry's long tradition of duplicity, that misleading AI listings will become the new norm.

Paradoxically, this may not be such a bad thing in the long run – for both agents and buyers. If AI images completely poison the image stream, the real estate aggregator will become a less and less useful tool for buying, selling, and finding homes. If this is the case, it is possible that the old system of agent exclusivity and personal business may come back into fashion because only offline interactions will be, ontologically speaking, effective. real. Given the fact that these aggregators have completely distorted the way we think about housing, housing and architecture in general, a world in which they are less influential or even useful could have interesting social and economic consequences.

From a cultural perspective, we're already seeing brands like lingerie company Aerie campaigning on refusal AI as a way to stand out from the noise, that is, they use humanity as a marketing ploy. (I find this quite sad on an existential level, but it's better than the alternative.) Like the body, a home is more than an image. It is physical, empirical and undeniably real. Perhaps we will begin to remember that the purpose of a home is not as an asset to be sold or as an object of varying degrees of verisimilitude, but as a place to live in, in which to live in the most vulnerable and human sense. It always mattered more than the number on the website. This will always have more meaning than photographs.