Amazon's Project PI detects product defects prior to shipment
Amazon has launched an advanced system named Project PI, also known as "Private Investigator." It employs generative AI and computer vision to spot defects in products before they are dispatched to customers. The system is engineered to detect damage, incorrect color, or size discrepancies in items. Products meant for customers undergo a scan in a tunnel, where the computer vision program checks for any issues and isolates defective items for further investigation.
Current operations and future expansion plans
Project PI is currently functional in several North American warehouses, with plans to extend its operations to more sites throughout 2024. This initiative succeeds Amazon's previous system, that identified frequently returned items to detect potential issues before customers place an order. Amazon stresses that averting a potentially "nightmarish" return process benefits not only customers and the company, but also aids in reducing carbon emissions.
Project PI's role in Amazon's second chance resell site
Items flagged by Project PI are examined by Amazon employees, who decide whether they should be sold at a reduced price on Amazon's Second Chance resell site or donated elsewhere. In addition, Amazon is developing a multimodal large language model AI tool, to understand why customers may be dissatisfied with their purchases. This tool scrutinizes customer feedback and images from Project PI to identify potential issues.
