Amazon begins to summarize customer reviews using generative AI

Over time, generative AI has proliferated in every sector of the tech industry. Due to the skills and powers of conversation and summarizing content, generative AI holds great potential. As per some recent pieces of information, Amazon has rolled out a new feature that uses AI to summarize user reviews. It generates a short summary of the advantages and disadvantages of a product that provides users with valuable insight and saves time.

Amazon declares that for months the feature has gone through testing and trial. Given the positive results, the feature is currently accessible to a few users in the US. The “AI-generated review summaries” will function similarly to the “Critics Consensus” and “Audience Says” summaries on websites like Rotten Tomatoes.

Let us understand this by using an example: a user is looking to buy a new LG smart TV. The user can benefit from the AI-generated summary feature that will indicate the positive and negative aspects of the products. Amazon is also integrating clickable tags that serve as navigational aids, emphasizing significant themes and common terms culled from the pool of customer reviews to further boost the AI summaries feature.

Fake user reviews and AI-generated summary feature

Well, this is such a good approach by the company however, it can be hampered by fake user reviews since AI will use the user reviews to highlight certain aspects of the products. A fake review can negatively impact the product. In response to this concern, the company states that the generative AI feature will rely on verified purchases for fetching information about a product.

Additionally, Amazon says it will use human investigators outfitted with cutting-edge fraud-detection techniques in dubious circumstances to examine and address instances of fake feedback. We still devote a considerable amount of resources to proactively battling bogus reviews. According to Amazon, to identify potential hazards, machine learning models examine a large number of data points, including account relationships, sign-in activity, review histories, and other indicators of anomalous behavior.