For Detecting Damaged Goods, Amazon Says AI Is 3 Times Better Than Humans
Amazon is using new artificial intelligence (AI) technology at two of its fulfillment centers to pinpoint and pull out damaged items before they ship to customers. It’s a move the Seattle tech giant believes will speed up picking and packing.
The company plans to roll out the technology at 10 more facilities in North America and Europe, according to a Wall Street Journal report. Amazon did not immediately confirm to Sourcing Journal which facilities are using the AI technology.
More from Sourcing Journal
Dr. Martens Reaches 1 Billion Pounds in Annual Revenue for First Time
Amazon, Hit With Driver Lawsuit, Braces for 1,000-Worker Walkout
According to the report, Amazon expects the technology to cut the number of damaged items that make it to the customer’s doorstep, accelerate picking and packing times, and eventually play a critical role in broadly automating fulfillment.
The AI technology comes into play when goods are picked for individual orders and placed into bins that move through an imaging station, where they are checked to confirm the right products have been selected.
Next, that imaging station evaluates whether any of the items aren’t in tip-top shape. A bin with a broken product is broken will move to a worker who takes a closer look. If everything looks fine, the order will be moved along to be packed and shipped out.
Amazon trained the AI using photos of undamaged items compared with damaged one, teaching the technology the difference so it can flag a product when it doesn’t look perfect, Christoph Schwerdtfeger, a software development manager at Amazon, told the WSJ.
Schwerdtfeger said that the company has found that the new AI algorithms are three times as effective at identifying damage as a warehouse worker.
Amazon, which moves about 8 billion packages each year, estimates that fewer than one in 1,000 items it handles is damaged.
The moves come as Amazon is overhauling its larger national U.S. fulfillment network, instead shifting to eight regional hubs to cut costs while expediting delivery.
According to CEO Andy Jassy’s annual shareholder letter, Amazon made significant internal changes across placement and logistics software, processes and physical operations to create the eight interconnected regions.
The company deployed advanced machine learning algorithms to predict what customers in various regions will need “so that we have the right inventory in the right regions at the right time,” said Jassy.
Jassy said Amazon is pleased with the preliminary results from its recently completed regional rollout. In the U.S., more than 76 percent of the products customers order are now from fulfillment centers within their region, according to the company.
The CEO said Amazon expects to see its fastest ever delivery speeds this year, with over 26 million shoppers ordering items for same-day delivery last quarter—up 50 percent from the same 2022 period.
Much of the wider debate about AI and automation in the workplace comes with concerns that the technology will replace human workers and lead to fewer jobs. The counterargument is that AI will simplify current jobs by freeing humans from conducting arduous, repetitive tasks.
It appears that AI could improve Amazon warehouse workers’ on-the-job experience.
While warehouse workers at Amazon are often responsible for checking goods for signs of wear and tear as they pick and pack goods, this can be time consuming given that most items are in perfectly good condition, Jeremy Wyatt, director of applied science at Amazon Robotics, told the WSJ.
Much of the time these warehouse workers often are required to meet company-set performance expectations measuring how many orders they handle per hour.
Amazon has denied having fixed quotas, saying that its expectations are based on multiple factors including how a team at a particular site is performing.
Lawmakers have criticized Amazon’s pace of work, with New York, California and Washington passing legislation to curtail the use of warehouses productivity quotas.
Although Amazon has faced a steady drumbeat of workplace safety criticism, which the Occupational Safety and Health Administration has already cited four times this year, it has long prioritized automation to aid workers.
The company recently opened its largest facility in New England, a 3.8 million square foot robotics-powered fulfillment center in Windsor, Conn. In April, Amazon officially opened a robotics fulfillment center in Fort Wayne, Ind., and inaugurated another automation-heavy facility in Papillion, Neb. to kick off the year.
