Artificial intelligence could upend the way we shop for loved ones around the holidays. It’s just not ready to do everything yet.
A third of people say they plan to use ChatGPT or some other generative AI while holiday shopping this year, according to a September analytics survey by Adobe. Those surveyed said they wanted AI to surface good deals, brand recommendations and gift alternatives.
These shoppers might be getting ahead of themselves. When you pose shopping and deal-hunting queries to the most popular AI chatbots, limitations often show up. “Hallucinations”—AI speak for inaccurate information—sometimes tell you something is on sale when it isn’t or recommend outdated products.
And this fledgling technology isn’t fully integrated into popular shopping sites like Walmart and Etsy. The ones that aim to use it have only just begun experimenting.
“Retailers are getting better but are nowhere close to where they want to be,” says Sharmila Chatterjee, academic head of enterprise management at the MIT Sloan School of Management.
AI could still help you this holiday shopping season—you just have to manage your expectations.
AI shopping assistants
Generative-AI tools like Microsoft’s Bing and Google’s Bard scrub the internet in real-time to find answers to queries. OpenAI’s ChatGPT, which is trained on older data, began offering real-time search responses in September, but you need the $20-a-month ChatGPT Plus plan.
You can ask these bots to scan sites to summarize reviews, offer price comparisons and recommend products. For instance, go to Bard and type in the prompt “Should I buy the new AirPods Pro?” The bot’s response may include a pros-and-cons list or notes about what users have said about the product. You can then follow up with questions about similar products and cheaper alternatives.
Where bots fall short is keeping up with the latest markdowns.
In Microsoft Copilot, formerly called Bing Chat, a search for “What’s the best quality portable photo printer I can buy right now?” produced a response with some promising options and links to articles. But while the bot said users could buy each of the printers on Amazon for $99, none was available for that price. A similar query in Bard linked to products that weren’t offering the sales the bot suggested.
For now, you should check the retailers for pricing and other details.
One site that is working to solve the deal-hunting problem with AI is BrickSeek. The price-comparison site, which helps shoppers find items before they sell out, launched its Deals page in September. The new page helps customers find the best markdowns on products across major retailers.
And while Google’s Bard might lack deal-hunting skills, the company has a new deals page that offers up-to-date information on sale items. You can activate it by typing “shop deals” in Search.
Smarter search
Online retail companies are also experimenting with generative AI to make search results more insightful.
At the beginning of the year, e-commerce platform Shopify—which says it partners with millions of retailers worldwide—started using OpenAI’s technology to power its in-app AI shopping assistant. In October, it brought an updated version to its website.
Users can treat the bot like they are talking to a person. You can ask it to help find toys for your music-loving niece, and it might show you child-friendly drums, guitars and more. The bot may ask you for specifics and refine its recommendations. Just bear in mind, it will only show items from those retailers Shopify supports.
If you want more options, Copilot in Microsoft Shopping works similarly across the web and offers additional information such as price history and comparisons. It began rolling out to users in mid-November.
Online marketplace Etsy is testing a similar AI-powered search bot. You would pose a query like “gifts for fall” and receive a chatty response with additional search suggestions. The chatbot is accompanied by a tailored results page that, in this example, would show product categories such as “cozy blankets” and “fall home décor.” Etsy didn’t say when this would be widely available.
Generative AI should be able to help retailers meet customers’ wants and needs without making them do much work, says Oded Netzer, a business professor at Columbia Business School whose research focuses on AI. He likens it to asking a friend for advice, though in this case the “friend” has its own motive.
“They want to sell products,” he says.
Other AI shopping tools
Last week, Google began rolling out generative-AI capabilities to make gift shopping easier. Searching “gifts for a 7-year-old who wants to be an inventor” produces all sorts of coding toys and other relevant ideas, with links to helpful articles. It is accessible to nonbusiness users in Chrome browsers and Google apps via Search Labs.
Also coming to Google’s Search Labs in December: an apparel-focused tool that lets users dream up what they are looking for using an AI image generator, which then points users to similar products that actually exist.
Amazon hasn’t added generative AI to its main search, but it did begin using it to summarize product reviews for shoppers in the U.S. The AI-generated text, which appears on product-detail pages in the app and website, highlights reviews from verified purchases.
Walmart is working on ways AI can enable open-ended searches. The company’s example, “unicorn-themed toddler birthday party,” could replace separate searches for multiple items, including “plates, streamers, party favors and more.” The retailer is also experimenting with a tool that blends generative AI with augmented reality to provide personalized, budget-specific interior-design assistance to customers designing a room.
When it comes to shopping with AI, there is huge potential, MIT’s Chatterjee says. But, she adds, we have yet to see “the proof of the pudding.”
—Dalvin Brown contributed to this article.
—For more WSJ Technology analysis, reviews, advice and headlines, sign up for our weekly newsletter.
Write to Cordilia James at cordilia.james@wsj.com
