While Amazon has long relied on artificial intelligence and machine learning to deliver targeted ads and personalized recommendations, it now seems determined to ride the wave of generative AI, the technology that powers ChatGPT.
Within its application, the e-commerce giant is testing generative AI to summarize the comments left by customers under certain products.
Summarize thousands of reviews to make it easier for shoppers
This new function, spotted by the media CNBC, provides insight into what customers liked or disliked about a product. It comes with a disclaimer that the summary is ” AI generated from customer reviews text “.
On the third-generation AirPods product page, for example, the AI summarizes the more than 4,000 user reviews. She explains that Apple’s headphones ” received positive feedback from customers regarding sound quality and battery life “. She indicates, however, that they received ” mixed reviews on performance, durability, fit, comfort and price “.
Amazon has confirmed that it is testing this feature, and one of its spokespersons clarified that it is investing ” significantly in generative AI » in all of the group’s activities. Using AI to summarize reviews could be useful for buyers, as there are sometimes many reviews under certain products. Nevertheless, technology, which tends to ” hallucinate “, still needs to be improved. Its formulations can also seem cumbersome and counter-intuitive.
Amazon goes for generative AI
This new initiative from Amazon is not surprising. In a letter to the company’s shareholders in April, its CEO Andy Jassy assured that generative AI and language models were able to ” transform and improve virtually every customer experience “. At the same time, the American giant is carrying out the largest wave of layoffs in its history.
Amazon is also considering integrating the technology into its site’s search tools. Last month, Dave Limp, senior vice president of devices and services, said that integrating generative AI with Alexa would be particularly relevant. In particular, this would allow users to interact with the virtual assistant in a more fluid and natural way.
Sources: CNBC, CNN
4