Artificial intelligence (AI) is integrating into people’s daily lives whether or not they are aware. One of the more visible examples of AI integration is the use of automated retail shopping in brick and mortar stores opening around the country.
In the last few years the world’s most successful online retailer, Amazon.com Inc., has launched cashier-less retail stores known as Amazon Go. The store’s “Just Walk Out” technology combines cameras, an app, and machine learning to automate the shopping experience. Shoppers enter the store and are greeted by an employee standing in front of an electronic gate with lanes. The employee prompts a first-time shopper to download the new Amazon Go app. The app connects to accounts they already use for online shopping and uploads the credit cards that are on file. Once they select their payment method, the app generates a key that allows them to scan through the gate.
Once a customer scans their key and walks through the gate, a virtual cart is created, and the sensors start tracking their movement throughout the store. The technology inside of the store is visible on the ceiling which is spotted with hundreds of cameras and sensors. The sensors monitor the motion of the customer as they walk up to a shelf and take an item. When they take an item, it is placed in their virtual cart. If the customer changes their mind and returns the item, then the store removes it from their virtual cart. Customer’s credit cards are charged when they walk back out of the gate, and then they receive an itemized receipt inside of the app.
This is all made possible through the sensors and machine learning that the store uses to monitor customers as they shop. The hundreds of sensors on the ceiling combine information gathered from camera lenses, infrared sensors, and motion sensors to determine which items are removed and which are replaced. Store clerks restock the shelves when they are notified by the store’s system that a quantity is running low. The cameras can track product labels, while the motion sensors monitor when that label moves. The infrared sensor determines when and where the customer is making contact with the product. All of this information is then synchronized to determine whether a customer has taken an item or not.
The system is highly accurate, but not perfect. Errors can occur when objects such as glass bottles with similar labels are placed directly next to each other. Grabbing a bottle of cola might register as root beer if the labels are similar enough. Maintenance and updates to the system allow the AI to continue learning about the differences between products. Amazon Go offers a no questions asked immediate refund through the app if the system made a mistake.
It is fair to consider that this level of automation could reduce employment opportunities in such stores. However, so far the system has only replaced cashiers. The store still requires clerks to stock the shelves, clean the restroom, and receive shipments. There are multiple associates on the sales floor at any given time, and always someone informing newcomers how the store works. Another position that is reduced but not totally unnecessary is security. Since everyone must scan in with an account that has a valid credit card attached, it is otherwise impossible to steal products from the store. Anything a customer takes off of the shelf will be monitored until it leaves the store and their card is charged. If someone has a negative balance, they will have to pay off the debt before entering the store again.
Right now, Amazon Go stores only sell food in the business districts of cities like San Francisco and Seattle. Selling food is a cheap way to test out an AI system that could integrate into larger retail stores with more expensive products. Being located around office buildings ensures a lot of foot traffic of people in a hurry. Shopping experiences are reduced to seconds rather than minutes, and products are paired together on the shelves to easily grab and go. The convenience of the experience will surely revolutionize brick and mortar stores in the years to come
Victoria Liset is strategic business & technology consultant to SMEs. She helps businesses improve their performance by using data more efficiently, and helping them to understand the implications of new technologies such as AI, Machine Learning, Big data, blockchain and IoT.