Pinterest is trying to get you to buy even more things by training your smartphone’s camera on the world around you.
Originally focused on helping people find and collect ideas about everything from couches to cakes in online pin boards, Pinterest has been steadily working to remake itself as a place to shop. A little over a year ago, it started adding “buy” buttons so users could more easily go from simply looking at items on its apps and website to actually purchasing them (see “EmTech Digital: Pinterest’s Bid to Reinvent Online Shopping”). The site, which has 100 million monthly users, now has 10 million buyable pins.
On Tuesday, the San Francisco startup previewed a deep-learning-assisted visual search feature that it hopes will encourage even more shopping. With it, you will eventually be able to snap photos of objects around you—a cool-looking T-shirt worn by someone who breezes past you, for instance—or even a whole room. Then you’ll see similar items that others have pinned to Pinterest, and if they have “buy” buttons, you can purchase them.
Software engineer Dmitry Kislyuk showed off a demo of the feature by using a Pinterest app on his iPhone to snap photos of items like a watch and a black rain boot; the app then showed a number of items on Pinterest that closely matched the ones in front of him. When he took a picture of a stuffed bunny that it couldn’t immediately identify, a bunch of bubbles popped up with words the app thought were related to the object, which he could tap to conduct a search.
Companies have long tried to marry real-world image recognition with online shopping—Amazon, for instance, has a feature in its mobile app that lets you take pictures of items that it will then attempt to find in its vast online store. But while such features can be helpful for finding, say, a specific DVD or a bottle of wine, it can be harder to use them to track down a succulent plant or a decorative bowl (see “AI Advances Make It Possible to Search, Shop with Images”).
Compared with Amazon’s offering, Kislyuk said, “what we’re trying to build is much more generic.”
Pinterest said its new form of visual search will be available in the coming months. It builds on another feature the company rolled out six months ago that lets you search in real time within images already on Pinterest; the company updated this feature as well on Tuesday, to make it more automated, and said that people have conducted 130 million visual searches per month since it was launched.
This new data poisoning tool lets artists fight back against generative AI
The tool, called Nightshade, messes up training data in ways that could cause serious damage to image-generating AI models.
The Biggest Questions: What is death?
New neuroscience is challenging our understanding of the dying process—bringing opportunities for the living.
Rogue superintelligence and merging with machines: Inside the mind of OpenAI’s chief scientist
An exclusive conversation with Ilya Sutskever on his fears for the future of AI and why they’ve made him change the focus of his life’s work.
How to fix the internet
If we want online discourse to improve, we need to move beyond the big platforms.
Get the latest updates from
MIT Technology Review
Discover special offers, top stories, upcoming events, and more.