“L O V E L Y Oakland!” the house description began. It went on to give a slew of details about the 1,484 square-foot home — light-filled, charming, Mediterranean-style, with a yard that “boasts lush front landscaping” — and finished by describing the “cozy fireplace” and “rustic-chic” pressed tin ceiling in the living room.
It looked like any number of the property descriptions you might see online in today’s red-hot US housing market. Typically, they’re written by humans. This one, however, was composed by artificial intelligence.
A Canadian startup called Listing AI is using cutting-edge AI to quickly churn out computer-generated descriptions that, like the one above, can be surprisingly compelling. All users need to do is give it some details about the home, and the AI does the rest.
The results still need work: The real-life Oakland, California, home that fits with the above description (which my family is currently selling) actually has a pressed tin ceiling in the dining room, rather than the living room, for instance. The descriptions Listing AI created for me are not nearly as specific or well-written as the one crafted by our (human) realtor. And I had to provide the website with a lot of information about different rooms and features of the house and the outdoor landscaping — a process that felt a bit like real-estate Mad Libs — before the website was able to come up with several different descriptions.
But the general coherence of the descriptions that Listing AI proposed within seconds of my submission provides yet another sign that AI is getting better at a task that was traditionally seen as uniquely human — and shows how people may be able to work with the technology, rather than fearing it may replace us.
It probably won’t do all the work of writing a house description for you, but according to Listing AI co-founder Mustafa Al-Hayali, that’s not the point. He hopes it will complete about 80% to 90% of the work for coming up with a home description, which may be completed by a realtor or a copy writer.
“I don’t believe it’s meant to replace a person when it comes to completing a task, but it’s supposed to make their job a whole lot easier,” Al-Hayali told CNN Business. “It can generate ideas you can use.”
There is a basement, but it’s not finished
Al-Hayali, whose day job is senior software developer at Canada-based e-commerce platform Shopify, said he and his co-founder (Corey Pollock, a Shopify senior product manager), came up with the idea for Listing AI after both became new homeowners in Toronto. As they each navigated that process, he said, they noticed property descriptions that were inaccurate or even copied from a previous time the property was up for sale. Al-Hayali bought his condo in March 2020 — just before Covid forced many people to start relying on virtual visits and other ways of learning about properties remotely, which made thorough descriptions that much more important.
They built the website over the past month or so and launched it publicly this week. Listing AI asks that a user provide all manner of data about the house, then software polishes it so that it can be better used by AI.
After that, the information is processed by GPT-3, an AI model from nonprofit research company OpenAI. GPT-3 was trained on text from billions of webpages so that it would be adept at responding to written prompts by generating everything from news articles to recipes to poetry. (Real estate descriptions are among the hundreds of applications developers have envisioned for the model, with varying levels of success.)
Users can sign up for a free trial that lets you generate several listings. If you want to generate unlimited descriptions it currently costs $9 per month or $84 per year (the website states these are launch prices; the regular prices will be $12 per month or $105 per year).
There were obvious errors and weird AI decisions in the listings the website composed for me, ranging from false statements (the house does not have a “finished basement”, though it does have a basement) to final sentences (“The living room”) that just trailed off. Listing AI also offers more social-media friendly versions of its home descriptions, which didn’t appear that much different to me but each came with a made-up asking price for the home (they ranged from about $600,000 to $2.4 million).
However, the overall tenor and style — and the fact that GPT-3 has no knowledge of the actual look or floor plan of the house — were surprisingly reflective of reality.
“This Mediterranean style house is perfect for you!” exclaimed one listing the website generated. It ended with a call to action: “Come and take a look today!”
After experimenting with Listing AI, I asked the realtor representing my family in the sale of my mother-in-law’s home, Scott Ward, what he thinks of this kind of tool. Ward, a realtor with Red Oak Realty in Oakland, had a few issues with Listing AI’s word choices — he’d never describe the Oakland house as “rustic-chic” he said, and he hates the word “boasts” in property descriptions.
But he agreed that there are plenty of poorly written home descriptions out there, and that automating the process could help some people. For example, it could be useful if people are not familiar with this kind of writing process or they’re selling a number of similar homes, such as tract housing.
Still, he said the assumption that AI, rather than a person, can be used to present a home to prospective buyers worries him.
“I think parts of this business can be certainly automated but there’s too much human touch still,” he said. “And I think that’s not necessarily a bad thing.”
Correction: An earlier version of this article misidentified Corey Pollock’s job title at Shopify. He is a senior product manager there.