May 15 (News On Japan) - AI-based search is rapidly transforming online shopping behavior, with users—especially younger generations—shifting away from traditional web searches and turning to tools like ChatGPT to find and purchase products.
This trend is particularly evident in Japan and the U.S., where major e-commerce platforms like Rakuten and Amazon are seeing a dramatic increase in traffic originating from AI searches.
According to a nationwide survey of men and women in their 20s to 60s, more than half still search for products on Rakuten or Amazon, while Google remains strong. However, the number of users who use AI tools like ChatGPT for product searches has reached 7%, with one in five people in their 20s using AI—a notable increase. AI researcher and EC consultant Tomoyuki Mochizuki, who participated in the survey analysis, pointed out that AI search offers unique advantages when product requirements are more abstract or nuanced—such as looking for a "stylish" portable power supply.
In one year, the number of users accessing Amazon via AI search grew 13-fold, while traffic to Rakuten rose 69-fold. The surge began around October last year, possibly coinciding with the adoption of newer AI models and enterprise use. On April 28th, OpenAI announced it would implement a shopping feature within ChatGPT, allowing users to browse product images, prices, and reviews in a tiled layout, similar to traditional e-commerce sites. In the U.S., users will soon be able to complete purchases directly through the interface, using linked accounts like Amazon.
This development has drawn significant attention from companies. While concerns remain about personal data and security, current implementations do not store credit card information within ChatGPT itself. Instead, they rely on external services like Amazon accounts. The AI system’s memory function, which recalls a user's preferences and behavior, enables increasingly personalized product recommendations, encouraging users to "teach" the AI over time.
OpenAI's strategy appears to be moving toward a broader platform model. Unlike conventional IT services that become more profitable as users increase, ChatGPT currently incurs massive electricity costs for each use due to its large neural network, reportedly running in hydro-powered data centers. This has led to unsustainable deficits, prompting OpenAI to monetize through e-commerce commissions and potentially future advertising.
While Google also integrates AI-generated overviews into its search results—implemented in Japan since August—these summaries do not yet appear for most shopping-related queries, likely to protect its advertising business. AI overviews have already led to reduced visibility for lower-ranked websites, raising alarm among businesses reliant on search exposure.
AI search also poses challenges in accuracy. There have been instances of incorrect information, such as misidentifying the current Japanese era, and researchers caution that AI-generated summaries cannot fully replace validated sources. University professors, for example, have noted a rise in students copying AI-overview text directly into assignments, sparking criticism of misuse.
Despite these flaws, AI search provides strong advantages in product discovery—especially for complex, time-consuming purchases or health-related queries. Its ability to interpret vague or abstract descriptions and filter through massive amounts of data in minutes makes it an efficient tool for shoppers seeking tailored results.
That said, physical products like books still present limitations. As Mochizuki noted, he prefers to buy books in brick-and-mortar stores where he can preview content directly—something AI search cannot replicate due to copyright restrictions. Ultimately, AI search is expected to coexist with other shopping tools such as social media platforms and video-based reviews, with different demographics and products favoring different discovery channels.
Consumers are advised to use AI search tools regularly to build more personalized experiences while remaining cautious about sharing sensitive personal data. As the technology evolves, companies and consumers alike will need to adapt to a new retail landscape shaped increasingly by AI.
Source: ABEMA
