Amazon Debuts Generative AI That Helps Sellers Write Product Descriptions

Many online product descriptions already read like they were written by AI. 

But the real danger is that AI produces false claims about products. JL 

Sarah Perez reports in Tech Crunch:

Amazon today introduced a new set of generative AI tools aimed at sellers which the retailer says will simplify the process of creating product listings. The retail giant claims these new capabilities are designed to help sellers generate “captivating product descriptions, titles, and listing details.”  eBay announced the launch of a generative AI tool that could generate product listings from photos. Shopify announced its own ChatGPT-like  for its merchants designed to interpret questions, and create content like blog posts and customer emails. The tools could contain mistakes if a human doesn’t review the output, creating incorrect product listings and descriptions

Amazon today introduced a new set of generative AI tools aimed at sellers which the retailer says will simplify the process of creating product listings. The retail giant claims these new capabilities are designed to help sellers generate “captivating product descriptions, titles, and listing details.”

Sellers will also be able to add to their existing product descriptions using AI, instead of having to start from scratch.

The AI tools were built using large language models, or LLMs, that were trained on large amounts of data. Though Amazon doesn’t specifically say, it seems that the retailer likely scoured its own listing data to train its machine learning models. Previously, Amazon had used machine learning and deep learning techniques to extract and enrich product information, but the new generation AI capabilities builds on that technology.

“With our new generative AI models, we can infer, improve, and enrich product knowledge at an unprecedented scale and with dramatic improvement in quality, performance, and efficiency” explained Robert Tekiela, vice president of Amazon Selection and Catalog Systems, in a statement. “Our models learn to infer product information through the diverse sources of information, latent knowledge, and logical reasoning that they learn. For example, they can infer a table is round if specifications list a diameter or infer the collar style of a shirt from its image.”

Amazon claims that its generative AI tools will help sellers save time and allow customers to find more complete product information, but there are some concerns around using generative AI models at such scale, given their ability to “hallucinate” — or create false information not based on real data.

The tools could also potentially contain other mistakes that aren’t caught if a human doesn’t review the output. And if the tools end up creating incorrect product listings and descriptions, Amazon could be held liable — particularly if it doesn’t disclose the listing was created using AI.

The Information previously reported Amazon was piloting generative AI tools for content, noting that the tool warns sellers to double-check the content to ensure it complies with Amazon’s listing guidelines. The company had declined to answer questions about the LLMM it was using for the new tool, the report had said.

Amazon is not the only retailer to turn to generative AI to make the process of creating product listings easier. eBay announced last week the launch of a generative AI tool that could generate product listings from photos. Earlier this summer, Shopify announced its own ChatGPT-like sidekick for its e-commerce merchants designed to understand and interpret questions or prompts related to business decision-making, and create content like blog posts, campaign ideas and customer emails, among other things.