Man vs. Machine - Robots in Retail
With his integrated shopping cart, the robot Budgee follows customers through the aisles of Walmart and carries their items to checkout. At Saturn, a German retailer for consumer electronics, his colleague Pepper is captivating with her adorable, large saucer eyes, but she is also useful, guiding customers to the product they are looking for. The lion’s share of the work done by these retail colleagues, who deliver consistently high performance, are always patient and never sick or overdue for a holiday, however, does not take place on the shop floor. Their main theatre of operations is logistics. They commission and palletise, but they can also solve more complex tasks. Technological advances have made robots more adaptable. They possess artificial intelligence (AI), and that makes them flexible — an inestimable advantage in comparison to the stage that precedes artificial intelligence, commonly described as automation.
Professor Michael Feindt of Blue Yonder goes considerably further in his analysis at the EHI Robotics4Retail conference in Cologne: “In terms of intelligence, robots can already run rings around us.” Artificial intelligence (AI = robots) is more reliable than human intelligence because people frequently use their oft-cited gut feeling and take decisions that are largely emotional and intuitive. A robot is entirely independent when it comes to this. In addition, it can take into account a practically endless number of influencing factors in its decision-marking and even integrate the ways in which they influence each other. This makes robots significantly more efficient than humans in many areas. They are unbeatable in retail when it comes to decisions involving processes that are frequently repeated, such as reordering or pricing. Prof. Feindt calculates that assuming roughly 40,000 items in approximately 1,000 supermarkets every day, this means 40 million decisions which can be made more quickly and more accurately by artificial intelligence than by a person. He illustrates this with an example.
A supermarket's rate of out-of-stock items was 7.5 percent. Initially, AI was used, but a human stock manager made corrections to its recommendations. The proportion of out-of-stock items dropped to five percent. Then the step involving correction by a human was left out. The surprising result was that the rate of out-of-stock items fell to 0.5 percent. This phenomenon is called cognitive bias and is what makes humans so prone to error: they have a tendency to make corrections to their judgements based on feelings. AI decisions are unbiased and therefore usually more correct.
One advantage of AI is that it can learn independently in a sense, allowing it to integrate a kind of learned experience. According to Prof. Feindt, this will soon make it possible for AI to not only reorder stock at the appropriate time and price items; in fact, these artificial colleagues will also be able to determine the right time for a sales promotion, for example.
This expert advises German retailers not to be so hesitant with new technologies, but rather to dare to try something new because companies such as Amazon and Google have long been using AI, and they are already at an advantage thanks to a big head start.