But Salakhutdinov says having a wealth of information about how users perform common and important tasks, like shopping, could be a crucial ingredient in keeping them on track. “The data is going to be very important,” he says.
send it
Of course, Amazon agents are likely more focused on helping customers find and buy what they need or want. A Rufus agent could notice when the next book in a series someone is reading is available and then automatically recommend it, add it to their cart or even buy it from them, says Rajiv Mehta, an Amazon vice president who works on conversational AI. shopping. “I might say, ‘We have one for you.’ We can ship it today and it will arrive at your doorstep tomorrow morning. Would you like that?’” Mehta says. He adds that Amazon is thinking about how to incorporate advertising into its recommendation model.
Chilimbi and Mehta say that eventually an agent might go on a shopping spree when a customer says, “I’m going on a camping trip, buy me everything I need.” An extreme, though not impossible, scenario would involve agents deciding for themselves when a customer needs something, and then purchasing it and shipping it to their door. “Maybe you could give him a quote,” Chilimbi says with a smile.
Amazon’s new AI-generated shopping guides, announced today at its Reinvent conference in Nashville and initially available on the company’s US website and mobile app, are a small step toward the ultimate vision of a super smart shopping assistant. Rufus LLM is used to automatically generate the type of information and knowledge that would take someone hours of online research to gather. “If you’re ever trying to shop in a category you’re not familiar with, it can take quite a while to understand the lay of the land, the different features available, and the different selections,” says Brett Canfield, a senior product. Amazon Personalization Team Manager.
Canfield showcased WIRED buying guides for TVs and headphones that highlighted important technical features, explanations of key terminology and, of course, recommendations on which products to buy. The underlying LLM has access to a vast corpus of product information, questions, customer reviews and comments, and users’ purchasing habits. “This is really only possible with generative AI,” says Canfield.
The new buying guides highlight the potential of generative AI in e-commerce, creating guides for product categories too specific to receive the usual treatment. “The ultimate hedge trimmers”, for example.
Guide Supplies
However, the guides also show how generative AI threatens to disrupt the economics of search and purchasing, while borrowing liberally from conventional publishers.
AI-generated search results now often offer product comparisons and reviews. This diverts traffic from outlets like WIRED, which make money by producing buying guides, reviews and other articles, although the AI results are produced using data extracted from such websites in the first place.
Canfield declines to say what additional training data was used to create the new AI shopping guide feature. (WIRED’s parent company, Condé Nast, partnered with OpenAIthe company behind ChatGPT, in August of this year).