In order to ensure that customers receive satisfactory products,AmazonUse your sharp weapon——"Project PI” (Detective Project). This combinationGenerative AI Systems that use AI and computer vision technology can identify damaged, wrong-color or wrong-size products before they are shipped to customers.
Here’s how it works: The incoming goods pass through a tunnel equipped with a scanner. A computer vision program (an AI technology that analyzes the content of an image) checks the goods for defects.If a problem is found, the system will separate the product., and conduct defect assessment, and check whether similar problems exist in other batches of goods in order to trace the source.
According to Amazon, Project PI is currently in use in multiple warehouses in the United States and will cover more sites this year. Last year, Amazon also launched another system that can mark frequently returned items to help customers avoid potential problem products before purchasing. These measures are all aimed at avoiding customers from falling into a "nightmare" return process, which is not only beneficial to customers, but also to Amazon itself and the environment (reducing carbon emissions).
Amazon said,Human reviewers will review the products flagged by Project PI, and decide whether to sell it in Amazon's unique "Second Chance" discount area or donate it to other organizations.
Amazon is also working on introducing a multimodal large language model to investigate the causes of customer dissatisfaction. The AI tool will analyze customer feedback and then combine it with images captured by Project PI and other data sources to find out where the problem lies. Amazon said the technology can help other sellers identify whether they accidentally mislabeled items.