Amazon Has Used AI to Avoid 2 Million Tons of Excess Packaging

Meghan Hall
·2 min read

Amazon is putting the “AI” in “mail.”

Its proprietary Package Decision Engine helps optimize packaging choices for more sustainable outcomes.

More from Sourcing Journal

The AI-powered model has aided Amazon in steering clear of over two million tons of excess packaging material in select markets since 2015, the company said. Now, it has begun training it for global use.

According to the company’s blog, items arriving at fulfillment centers get photographed in a computer vision tunnel to identify dimensions, attributes and more. That image can also flag defects, which can help ensure faulty items don’t make it past the warehouse.

From there, the model uses text-based data on every item to make a prediction on the most viable—yet simultaneously efficient—type of packaging to use. Because it can identify from the data what the product is and has been trained on customer feedback, keywords and prior successful shipments, it not only determines the best size for the mailer or box, but also pinpoints whether a product needs additional protection or cushioning.

For example, apparel items like jeans or T-shirts may work without issue in a low-cushioning mailer, but more fragile items like a watch with a glass face may require a heftier option.

Prior to the model, Amazon associates would individually test packaging options for products, but according to the company’s blog, “that sustainability work was impossible to scale through human efforts alone.”

Kayla Fenton, senior manager of technology products with Amazon’s packaging innovation team, said the tool has shown promising results.

“We wanted the ability to quickly identify the most efficient packaging option for each item, while also predicting how safely each product would ship,” Fenton said in a statement. “The use of AI through the Package Decision Engine has allowed us to advance our packaging efficiency work at scale quickly, and it has worked so well that we’re implementing this technology across Amazon’s broader global footprint.”

But even as the e-commerce giant works to create less waste, an Oceana report from earlier this month shows its U.S. plastic footprint looks dicey at best.

Amazon already uses the Package Decision Engine in North America and Europe, but will begin rolling it out in India, Australia and Japan shortly. To do so, the company “exposing it to new languages, unique packaging types and items sold in different countries,” it said in the blog.

The packaging selector isn’t Amazon’s only sustainability-focused AI investment.

It uses an AI model to predict the correct size when customers order clothing, in the hope that it decreases return rates. It also leverages the technology to flag defective products, ensuring a human reroutes the product to be sold on a secondary market, donated or otherwise used. According to the company, AI can identify damaged goods with three times the efficacy of humans.