I Tried the Meta AI Product Search Service and I Won’t Be Using It Again.
I usually try to avoid using Meta AI , but today I gave it a try. According to Bloomberg , Meta AI is currently testing a shopping assistant. The idea is to compete with similar services from AI platforms like ChatGPT or Gemini, where users leverage generative AI to search the web for product recommendations. While this sounds good in theory, in practice, Meta AI’s shopping assistant is a complete failure, even if it’s just a test.
Meta AI’s buy offers are useless.
When I tried accessing the Meta AI web app without logging in, I didn’t see the shopping assistant. But once I logged into my Meta account, several additional options appeared, including “Shopping.” Clicking this option in the model selector reveals a new “Shopping Search” option, along with “Thoughts” and “Quick Search.” While you can enter your own search query in the search bar, a series of suggestions also appears below. Currently, I see queries like “Help me find a signature fragrance,” “Show me mid-century living room furniture,” or “Find a unique gift for my friend.” Finally, the AI is designed to help you spend money when you can’t decide what to buy.
That last question really intrigued me. Meta AI doesn’t know my friend, and yet it’s going to find me a “unique” gift they’ll love? I needed to see what the bot would suggest. After Meta AI thought for a bit (about 11.6 seconds, it showed me), the bot actually confirmed my suspicions by asking for more information about my friend. It wanted details like hobbies and interests, budget, and any special memories I could include in the gift. You’d think some gift ideas would have come to me just by thinking about these questions, but I needed the AI’s help, so I offered the following fictitious answers to test it: “Their hobbies and interests are music, cycling, and going to the movies. My budget is $100. We went on vacation to Hawaii and missed our flight home.”
The artificial intelligence thought for 45.8 seconds before returning with its “unique” gift ideas. Its favorite was something that combined my “friend’s” interest in music and our fictional Hawaiian vacation: a ukulele. A truly unique gift. Meta AI reasoned, “A ukulele starter kit is the perfect way to bring island vibes into your home. The Kala Learn to Play Ukulele Concert Starter Kit fits your budget and includes everything you need to get started, including lessons and a tuner. It’s a great hobby for a music lover and a direct connection to Hawaii.” Other unique gifts included three different types of portable speakers. Thank you.
“For the movie buff,” Meta AI suggested a poster with a list of 100 movies everyone should see. Or, as a “subtle nod to our trip,” a poster for a film shot in Hawaii. Admittedly, this one is a bit more original, although I’d never buy one. But after scrolling past the bicycle accessories section, I stumbled upon a real find:
Meta AI won’t be of much help if you search for products yourself.
After failing to find the perfect gift for my imaginary friend, I tried using a custom query: “I need a new sofa for my living room.” This proved slightly more helpful. The bot returned a list of five different sofas, each with a description of the design and brand, along with a carousel of 12 sofas at various prices. To narrow my search, the bot asked me a few questions about my home and interests, such as the size of my living room and the style I was looking for. I sent the bot the following: “My living room is 200 square feet. I need a modern leather sofa under $2,000.” This returned what I thought were a suitable list of sofas, each of which matched my budget and style. I also learned about some new brands I hadn’t heard of before, like Article and Poly & Bark.
Finally, I turned to Meta AI for help finding a new MacBook. I told it I needed something fast, but under $1,200. It did some research and came up with three decent options: a 13-inch MacBook Air with an M4 processor for $999, a 15-inch MacBook Air with an M4 processor for $1,199, or a MacBook Air with an M3 processor and extra storage for $1,030. There’s nothing wrong with these suggestions, but I noticed two problems with the results. First, the link to the third MacBook Air recommendation didn’t lead to an ad, but to the homepage of Abed Tahan, a store in Lebanon. A quick search on the site returned results for the MacBook Air with an M3 processor, which I was interested in, but it was more expensive than Meta AI had indicated, and the store doesn’t ship to the US. Second, I asked Meta AI if these results were the latest models, and it confirmed they were—despite Apple announcing new MacBooks with M5 processors this morning . If the bot were working with a limited knowledge base, that would be one thing, but it searches the internet with every request. It needs to be able to find this information.
While the Meta AI shopping assistant has obvious advantages, such as its ability to find furniture that matches the size and style of a specific room, I’m almost certain I won’t use it. The fact that some product links don’t work and that it can’t reliably provide the most relevant products on the market defeats the purpose of using a shopping tool. I’ll stick to my usual search method: traditional search combined with real-world user experience.