Artificial intelligence in retail today

In recent years, hardly any other topic has animated the world of retail as much as the development of artificial intelligence and its possible uses. We will show you what is already possible today.

Before the coronavirus pandemic dominated current events and the media, the term artificial intelligence (AI) appeared regularly in the (specialist) media. Without AI, there is no Industry 4.0. AI is at the core of autonomous vehicles and even household appliances will become increasingly “smart” with AI. It is time to take stock from the point of view of the retail industry.

Not everything “AI” is “intelligent”

One of the major misconceptions related to “artificial intelligence” comes from the use of the term “intelligence”. It is very human to anthropomorphise, that is, to believe that IT systems act and “think” like us. This misconception is a consequence of this imprecise term, as AI encompasses a variety of methods and approaches. These include machine learning and pattern recognition, which form the basis for the majority of current AI solutions.

With machine learning (ML), systems actually create hypotheses on their own and propose solutions autonomously. If a hypothesis works, it is adopted as “knowledge”. For example: Based on the current weather forecast, the AI suggests products that match the weather to customers in a particular region. If these products are then actually purchased more often, the AI applies this knowledge to future events or other regions.

We already encounter pattern recognition in everyday life when we communicate with chatbots or use voice assistants or auto-complete on our smartphones.

Technology has made enormous progress in both areas in recent years. But machines that think and act like humans do not exist yet.

AI in retail: An (incomplete) list of examples

Without a doubt, some of the most spectacular developments in AI that are also visible to customers, are the automated shops that operate without a cash desk and (apparently) without staff. While Amazon relies on AI, cameras and sensors for its “Go” concept, Aifi depends entirely on machine learning and camera technology, and recently opened its largest store (400 square meters) using this technology.

Somewhat less exciting are smart shopping trolleys, such as those from Caper, which are equipped with sensors and cameras, and recognize the products taken by customers when they are placed in the shopping cart. This simplifies self-checkout.

At H&M in New York, voice-controlled mirrors give advice directly in the store, offering customers alternatives to the selected items or, as in the case of Asos, a virtual catwalk that customers can use in their own home.

Systems for dynamic pricing may seem a little old school, but almost nothing works without AI here other than a simple re-pricer. Technologies that are already used very often in online retail are slowly conquering physical stores, for example at MediaMarkt-Saturn.

At physical stores, AI systems are optimizing personnel planning or dealing with energy management. For example, “energyControl” at Breuninger uses predictive AI to reduce energy costs and CO₂ emissions.

One example of pattern recognition is voice commerce via voice assistants or in the form of chatbots on retailer websites. The “LiA” chatbot, for example, is the direct line to Lidl. At Hit Sütterlin in Aachen, customers ask the Alexa voice assistant about special offers.

AI also helps with range design, so Zentrada works with AI using self-learning algorithms to provide associated smaller retailers with a better range of products.

Visual product searches, such as those that Amazon uses in its app, are another example of image recognition. Customers take a picture of a product using their smartphone's camera, and the AI attempts to find precisely that article (or a very similar one).

Finally, we shouldn't forget digital size guides, which use customer information and, if available, historical order data to determine the best fit. This keeps customers happy and reduces return rates. One example is the Fit Finder from Otto. For example, British fashion retailer Asos uses AI to create individual product recommendations, including fit, based on customer data.

When using machine learning and pattern recognition, AI systems also have their advantages at the checkout. When detecting and preventing fraud, AI systems, such as the Hybright system from CRIF Bürgel, can make decisions faster and more precisely than humans and use the data to control the payment mix.

US retailer Walmart is using AI to optimize picking of ordered goods as well as delivery from its warehouses and logistics hubs. AI optimizes vehicle utilization and minimizes travel distances. This is good for costs, but also for the environment.

Is AI going mainstream?

The advances in AI technology are impressive. And AI has already found its way into many areas of our lives without us always being aware of it. But we shouldn't assume because of this that AI is now also becoming mainstream in retail. The examples given are all solutions that are highly tailored to the respective situation of the company. With a few exceptions (e.g. chatbots or dynamic pricing), there are no off-the-shelf AI solutions. The systems often still have to be trained and set up individually which incurs costs.

We are still a long way from having AI solutions for the local corner shop. But it is certainly an increasingly attractive area for larger companies.