Zillow has big plans to make house hunting smarter with AI
Real estate is the next area AI is conquering, and first up on the battlefield is Zillow.
When Zillow’s chief marketing officer, Jeremy Wacksman, joined the company in 2009 (which, with the U.S. economy mired in the housing crisis, he says was a “fabulous time to join a startup in the real-estate space”), he was determined to incorporate artificial intelligence into the platform.
Currently, the site is undergoing what Wacksman describes as “the evolution from a search box to an assistant.” The ideas is to transform Zillow from a simple real-estate search engine to a tool that understands you. It’s doing this in a few ways:
First, it’s building AI into the search engine, in the form of a new feature called Personalized Sort, expected to roll out later this year. Details are few right now, but the idea is Personalized Sort will learn the types of criteria you’re looking for and recommend homes based on that.
The site already does this with in-home amenities to a certain extent. If you’ve been looking at a lot of apartments recently with, say, in-unit laundry, Zillow will recommend other homes with that feature.
Personalized Sort will take that idea to the next level. It’ll learn not just what you’re looking for in terms of a home, but apply those preferences to regions as well.
For example, if you tend to search in areas around a certain school district, movie theater, or other amenities, Zillow will be able to recommend additional homes to you that meet those criteria but may not be in the immediate vicinity.
The AI will take into account your taste in decor as well. The idea here is to take into account the interior photos of homes you’re looking at and which ones you seem to like. Then it’ll take into account what the kitchens, living rooms, and bedrooms look like when making recommendations. This should roll out in the first half of 2018.
“People now expect services to be smart, to be personalized,” Wacksman says. “We’re doing a lot of work with personalization, about how to recommend the right things to serve up and suggest to you.”
AI will also make an appearance in the upcoming rollout of Zillow’s 3D Home app, a Google Street View-like service for viewing the inside of a house.
The app allows anyone to take 360 photos of their property — no fancy 3D camera required. All you need to do is shoot a set of video files of the interior, which the app will stitch together by using AI to find the boundaries and areas of overlap. It then creates a fully immersive 3D image of the home that viewers can browse on a phone, or, if they have a VR headset, walk around in.
Zillow is currently testing 3D Home in Phoenix, but will release it in more locations throughout 2018.
Checking rental listings Zillow typically takes me 25 minutes or more. I have to click through every home in the immediate area I’m interested in to make sure I haven’t missed anything. I zoom in and out on dozens of blurry interior photos to compare minutia.
If Zillow’s AI delivers on the promise, that task won’t be as much of a chore. The recommendations should be guiding me to the homes that I’d like best, and the 3D imagery will be a better first look than just browsing still photos.
Still, there’s reason to be skeptical. It’s entirely possible, even likely, that Zillow’s algorithms may not be as nuanced as they need to be. Think about Facebook’s News Feed, where, for a time, watching one Tasty video would turn my entire feed into Tasty videos for months on end. And teaching a neural network what exactly constitutes a “good” kitchen could be a long process.
The payoff may be worth such bumps in the road, though. At the end of the day, Zillow users are parsing a massive database to make a decision that could impact the rest of their lives, and that process is getting harder every time more data is added. Most will take any help they can get — even if that comes in the form of an AI.