Amazon is leveraging generative AI to enhance same-day shipping by optimizing robot operations and improving delivery routes.

For years, has set the standard for package delivery. When Prime launched in 2005, two-day shipping was unheard of. By 2019, one-day shipping was standard for millions of items. Now, the retail giant is using generative AI to drive more same-day shipping.

Amazon is employing technology to enhance delivery routes, develop more intelligent warehouse robots, create ergonomic work environments for employees, and accurately forecast where to place new products, according to Steve Armato, Amazon's vice president of transportation technology and services.

Amazon is utilizing generative AI to boost the percentage of Prime orders delivered the same day or next day in the top 60 metropolitan areas in the U.S.

According to Armato, who was interviewed at the warehouse, although it may seem minor, placing one additional product in the correct location can significantly reduce the shipping distance when ordering.

In 2020, Amazon started employing transformer architecture, which is the foundation of generative AI, to create models for demand forecasting and supply chain optimization.

"The next major technological advancement is generative AI, as stated by Armato. It's impressive and we are already utilizing it in practical ways throughout our operations."

While generative AI may bring some benefits to e-commerce giant Amazon, there are also concerns about its high-energy needs and its potential to replace human workers with robots, according to analysts.

Robots and new roles

According to the company, the number of Amazon warehouse robots increased from 350,000 in 2021 to over 750,000 in 2023.

In 2022, Amazon introduced AI transformer models to its warehouse delivery robots, allowing them to move around each other more intelligently. CNBC observed hundreds of them moving in a coordinated grid in the warehouse, which Armato refers to as "the dance floor."

Armato suggested that some of the two-day deliveries could be postponed so that the robot with a next-day delivery could proceed first and take a direct path to its destination.

Amazon's next generation of drive units, called Proteus, are fully autonomous, the company said.

Armato stated that they are employing generative AI and computer vision to evade obstacles and locate the optimal stopping point.

In August, Amazon struck a deal with AI startup Covariant, hiring its founders and licensing its models that enable robots to handle a broader range of physical objects. Additionally, Amazon is developing a bipedal robot called Digit that can grasp and handle items in a humanoid manner as part of its AI strategy.

Amazon's "Robin" arms, equipped with computer vision, can determine the appropriate pressure to use when handling various package shapes and sizes. These arms are trained by generative AI to handle products they've never seen before based on data from similar products in Amazon's vast catalog.

Amazon's AI is more accurate than humans in identifying damaged products, the company claims.

Balancing the introduction of more robotics with generative AI and maintaining human workers is a challenge for Amazon, according to Tom Forte, senior equity analyst at the Maxim Group.

Forte posed the question of how to implement automation to enhance efficiency and control labor costs while also ensuring that it complements the use of humans and does not replace them.

By implementing robots, Amazon is shifting the workload from employees and creating new roles, Armato stated. The company plans to invest $1.2 billion to train more than 300,000 employees in generative AI and robotics by 2025.

"If the robot breaks down, someone must maintain it. If something is dropped on the dance floor, there is a process and special training to clean it up. This results in new categories of jobs, some of which have a higher earnings potential."

In recent years, Amazon has faced criticism regarding its workplace injury record, including federal safety violations and a Senate investigation that concluded that Prime Day, Amazon's annual sale, contributed significantly to worker injuries. Despite this, Amazon has challenged the citations and argued that the report fails to acknowledge the advancements it has made.

Amazon's robots are reducing the amount of walking and bending required for employees by moving tall bins of items to workstations, according to Armato. AI is also decreasing the need for workers to reach and bend.

""By placing our faster-selling products at waist height on the shelves, we can improve our algorithmic approach, as this is the ergonomic power zone," Armato stated."

Predicting orders and routes

In the first quarter of 2024, Amazon delivered over 2 billion items the same day or next day with the help of robots and workers, as per the company's statement.

Amazon has employed algorithms to forecast inventory requirements, and now the company is leveraging generative AI to determine the optimal locations for items that have not been previously sold.

Placing a product in the right place before clicking buy reduces the distance it travels, resulting in a win for both speed and sustainability, according to Armato.

Amazon Web Services has data centers filled with servers running AI workloads that give the company an advantage over its retail competitors because it can train its AI in-house. As an early online-only retailer, Amazon got a head start on collecting mass aggregate data on shopping behavior and delivery logistics. Amazon is now leveraging that vast data trove to develop AI models for use in a wide range of applications, including supply chain optimization and warehouse robotics, according to the company.

Sucharita Kodali, retail analyst at Forrester Research, stated that while others have their own data, they approach things differently and have older systems.

The long-term eco-friendliness of generative AI is uncertain due to its carbon-intensive training and running processes, which are projected to consume as much power annually as Sweden or the Netherlands by 2027.

Amazon's 2019 commitment to achieve net-zero carbon emissions by 2040 is in conflict with its current practices.

The company claims that the use of AI is helping reduce the carbon footprint of package delivery. Amazon is using more than 20 machine learning models to improve mapping for its vast network of 390,000 delivery drivers, predicting road closures and choosing more efficient routes, the company said.

Amazon has also introduced generative AI to aid its sellers and customers beyond its warehouses.

The Amazon Personalize AI tool generates hyper-personalized product recommendations for the company, while sellers can use generative AI to write highly targeted product descriptions and generate images of their products in different "seasonal and lifestyle" settings.

In 2023, Amazon started featuring AI-generated summaries of product reviews on its website, and in February, the company introduced Rufus, a conversational shopping assistant powered by generative AI.

Amazon has invested $4 billion in AI startup Anthropic, which creates chatbot Claude, a competitor to OpenAI's ChatGPT. Additionally, Amazon produces its own AI-focused microchips and generative AI tools for developers, which it also uses in its operations.

The profitability of Amazon's significant investment in generative AI is uncertain.

"Kodali stated that he has not observed significant growth in any retail business as a result of generative AI, including Amazon. He believes that Amazon's greatest impact has been due to earlier investments, rather than recent ones."

Watch the video for more on how Amazon is using AI.