Michael Anthony wasn’t immediately certain, but the child’s bunk bed quickly revealed the anomaly. The design featuring two intersecting stairs was clearly an unusual choice, signaling a peculiar error often associated with AI-generated images.
The 28-year-old software engineer recounted to VICE, “I had a feeling as I scrolled through the images, and suddenly something felt off.” Initially attributing the oddity in the Rightmove listing to an overly enthusiastic estate agent, Michael’s perspective shifted when online observers suggested AI involvement after he shared some images on the Spotted on Rightmove platform. Reflecting on the situation, he remarked, “Honestly, I found it quite impressive.” Indeed, AI technology has made significant strides in the past year, hasn’t it?
Michael stumbled upon a growing trend: real estate agencies leveraging AI for listing photos. Agents are increasingly enhancing property advertisements with AI-generated enhancements, transforming dreary rental spaces into IKEA catalog-worthy settings resembling a surreal version of The Sims.
Known as online staging, the practice of digitally adding furniture to real estate photos has been in existence for some time. However, recent advancements in generative AI tools have eliminated the need for costly design professionals or subpar Photoshop skills among agents, enabling them to effortlessly enhance home listings.
While public information on this emerging trend is scarce, there is a noticeable uptick in AI-powered advertisements on real estate platforms like Rightmove and Zoopla. Industry insiders informed VICE that as the costs associated with AI decrease, its adoption is expected to become more widespread. One of the largest real estate agencies in the UK disclosed that they are already offering third-party AI enhancement tools to their clientele.
Despite some agents experimenting with popular AI image generators like Dall-e and Midjourney to “improve” property pictures, the outcomes can be quite outlandish. For instance, AI may envision bookcases in every corner of a living space and position a stove at a height suitable for a child, as seen in a peculiar image of a kid-sized stove in a home. Companies such as Virtual Staging AI, ModelProp, and Gepetto, which describes itself as “Pinterest on pills,” are now utilizing this technology to create customized virtual furnishings.
In a real-world application that adds tangible value, Michael Bonacina, CEO of Virtual Staging AI, emphasized the convenience of generating captivating images online. For a subscription fee of $12 (£9.50) per month, estate agents and landlords can access the image generation app. Since its launch in March of last year, the US-based company boasts 3,000 paid subscribers, including UK-based agents, generating a staggering 500,000 images monthly.
Navigating through the unsettling array of property listings on Rightmove can be daunting, with images potentially distorted by wide-angle lenses or excessive editing, or even staged to a certain extent. Some UK listings feature the work of Virtual Staging AI, evident from the company’s name included in the filenames of edited photos. Detecting AI-generated interior design can often involve scrutinizing for garbled text, irregular textures, skewed angles, and stray pixels, although advancements in technology are gradually refining these telltale signs.
While Virtual Staging AI does not automatically append a logo or disclaimer to denote AI use, Bonacina cautioned that the absence of transparency could have adverse effects on prospective buyers. With the housing market’s competitiveness leading some tenants to make sight-unseen purchases, clarity regarding the staged nature of images becomes crucial.
Another concern lies in the potential for AI to misrepresent room sizes by furnishing spaces with items that are not to scale, creating a misleading impression of spaciousness. Omar Beg, managing director of McBryer Beg, a London-based real estate firm utilizing AI for ad enhancements, highlighted how AI-staged areas may inaccurately portray a property’s true potential, presenting unrealistic configurations that deviate from practical living arrangements.
Agents often select images that showcase rooms in their best light, even if the setup is impractical for actual occupancy, as AI staging services typically generate multiple design renders for each room. While visually appealing, these staged settings may not align with the reality of the living space, potentially leading to misaligned expectations for prospective tenants or buyers.
Acknowledging occasional errors in the past, Bonacina affirmed the continuous improvement in AI accuracy, expressing confidence in achieving practical results in the vast majority of scenarios in the near future.
The escalating use of AI in real estate, whether through online staging or other applications, has raised concerns among advocacy groups like Renters Union Acorn. Beyond the realm of false furnishings, AI algorithms are increasingly influencing real estate ad descriptions, with reports of landlords manipulating rental prices using AI-driven strategies, exacerbating affordability challenges for tenants.
In a landscape where affordable housing remains scarce, unethical practices by landlords and agents, whether through deceptive AI imagery or price manipulation, exploit the vulnerability of individuals seeking a place to call home. Nick Ballard, head organizer of Acorn, emphasized the need for transparency and ethical considerations in real estate practices to protect tenants from misleading representations and unfair competition over rental prices.
