Amazon Dreams of AI Agents Doing the Shopping for You

But Salakhutdinov says having more information about how users perform common and important tasks like shopping can be a key ingredient in keeping them on track. He says: “The details will be very important.

Send It

Amazon agents, of course, may be more focused on helping customers find and buy whatever they need or want. A Rufus agent can see when the next book in a series a person is reading is available and automatically recommend it, add it to your cart, or buy it for themselves, said Rajiv Mehta, Amazon’s vice president of AI conversations. shopping. It might say, ‘We have something you bought. We can ship it today, and it will arrive at your door tomorrow morning. Would you like that?’” Mehta said. He added that Amazon is considering how advertising can be incorporated into its recommendation model.

Chilimbi and Mehta say that in the end, the agent may go on a shopping spree when the customer says, “I’m going to go on a camping trip, and buy everything I need.” An extreme scenario, although unlikely, would involve agents deciding for themselves when a customer needs something, then buying it and having it delivered to their door. “Maybe you can give me a budget,” said Chilimbi with a smile.

Amazon’s new AI-generated shopping guides, announced at its Reinvent conference in Nashville today and available initially on the company’s US website and app, are a small step toward the ultimate vision of a super-intelligent shopping assistant. Rufus LLM is used to automatically generate the kind of information and data that would take someone hours of internet research to gather. “If you ever try to shop in an unfamiliar category, it can take time to understand the area, the different features available, and the different options to choose from,” says Brett Canfield, senior product. manager in the Amazon personalization team.

Canfield presented WIRED’s buying guides for televisions and earbuds that noted key technical features, definitions of key terms, and, of course, recommendations on which products to buy. A basic LLM has access to a wealth of product information, customer questions, reviews, and feedback, as well as the purchasing habits of users. “This is only possible with productive AI,” Canfield said.

New shopping guides highlight the generative power of AI in ecommerce, creating guides for product categories that are too unique to receive treatment. “Direct hedge trimmers,” for example.

Guidance Resources

The guidelines, however, show how productive AI threatens to boost the search and shopping economy while lending itself freely to mainstream publishers.

AI-generated search results often now provide comparisons and product reviews. This disrupts traffic to stores, like WIRED, that make money by producing shopping guides, reviews, and other articles, even if the AI ​​results are generated using data extracted from such websites in the first place.

Canfield declined to say what additional training data was used to develop the new AI shopping guide feature. (WIRED’s parent company, Condé Nast, entered into a partnership with OpenAI, the company behind ChatGPT, in August of this year.)