Amazon experiments with using AI to sum up product reviews

The lawmakers trying to mitigate risk from AI Tory MP Michelle Rempel Garner and independent Sen. Colin Deacon discuss how they're trying to mitigate AI’s potential risk to Canadians.

Amazon is experimenting with using artificial intelligence to sum up customer feedback about products on the site, with the potential to cut down on the time shoppers spend sifting through reviews before making a purchase.

On the Amazon product page for Apple’s third-generation AirPods, for example, the AI feature now sums up the more than 4,000 user ratings to note that the wireless headphones “have received positive feedback from customers regarding their sound quality and battery life.” But, it adds, “mixed opinions were also expressed about the performance, durability, fit, comfort, and value of the headphones.”

The summary features the disclaimer: “AI-generated from the text of customer reviews.”

“We are significantly investing in generative AI across all of our businesses,” Amazon said in a statement to CNN on Monday, referring to the technology that underpins services such as ChatGPT.

The effort, first reported by CNBC, marks Amazon’s latest attempt to incorporate generative AI into its services and has the potential to help customers quickly determine the pros and cons of various products. But there are limits.

For starters, the AI wording is not always intuitive. In the AirPods review, for example, the blurb says “all customers who mentioned stability had a negative opinion about it.”

As with other generative AI tools, which are trained on vast troves of data online to come up with responses, there are also concerns about tone, accuracy and its potential to “hallucinate” details.

“Given that generative AI is based on probability, mistakes are possible … and summaries may not be an accurate reflection of customer reviews,” said Reece Hayden, a senior analyst at market research firm ABI Research. “The possibility of hallucinations will be a worry for customers and merchants.”

Hayden also questions whether the tool will be able to decipher fraudulent or bot-created reviews. “These reviews will be treated equally and therefore the summary may reflect fake, non-customer reviews,” Hayden said. (Amazon didn’t immediately respond to a request for comment on this possibility.)

Amazon isn’t the only e-commerce company blending generative AI into the shopping experience. Some companies such as Shopify and Instacart are using the technology to help inform customers’ shopping decisions. Meanwhile, eBay recently rolled out an AI tool to help sellers generate product listing descriptions.

Amazon CEO Andy Jassy said in a letter to shareholders in April that the company remains focused on “investing heavily” in the technology “across all of our consumer, seller, brand, and creator experiences.” The company is also reportedly working on adding ChatGPT-like search capabilities for its e-commerce store, and it’s rumoured to be planning to use generative AI to bring conversational language to a home robot.

Last month, Dave Limp, senior VP of devices and services, told CNN there is great interest in bringing generative AI to virtual assistant Alexa, so users can interact with the technology in a more fluid, natural way.