Shoptalk and South by Southwest ran consecutively this year. In spending time at both, certain correlations stood out. The implementation of artificial intelligence in retail was examined in detail at Shoptalk, whereas its role in our overall economic activity and lives was an overarching theme in Austin at the SXSW Interactive Festival.
The Term Artificial Intelligence is Becoming a Catch-All
Before we go any further, it might be useful to set a baseline for AI. A simple definition of artificial intelligence according to futurist, professor, and author Amy Webb is, “In its most basic form, artificial intelligence is a system that makes autonomous decisions. The tasks AI performs duplicate or mimic acts of human intelligence, like recognizing sounds and objects, solving problems, understanding language and using strategy to meet goals.” With the explosion of computational power, these tools, first developed in the 1940s, have developed exponentially, and have quietly been woven into our daily activities. While typing the Amy Webb quote above, artificial intelligence, which informs my software, kicked in and highlighted a spelling error. Consider the magic behind Spotify, your Netflix queue or home voice assistant, and you will quickly recognize that beyond business applications, artificial intelligence lives with us, and we may have become so accustomed to the efficiencies it offers that AI is more than the new invisible hand, it has become our collective invisible brain.
In retail we see this brain at work in the utilization of predictive customer preferences, inventory efficiencies and supply chain optimization. Beyond efficiencies, artificial intelligence is foundational to subscription retail businesses like StitchFix and critical to most fast-fashion brands.
Humans and AI- Working Hand in Hand
Mike Smith, President and Chief Operating Officer of StitchFix, was at Shoptalk discussing their secret sauce, which is composed primarily of two ingredients: robots and humans. StitchFix creates its individualized “Fixes” or boxes using a combination of artificial intelligence and stylists’ knowledge to provide a customized selection for their subscribers. In turn, StitchFix subscribers provide an abundance of data to the company as they respond to the products in every box, which only improves the AI’s predictive power. StitchFix also offers an online and app game called Style Shuffle. Smith described Style Shuffle as a swipe right, swipe left exercise that gives the company reams of data on what people would choose to wear. According to a recent article in Fast Company, Style Shuffle led to a new initiative called Latent Style. Latent Style is a sorting system for “fashion clusters.” A flouncy-dress or clean-lines cluster are two examples. The predictive system then informs the individual stylists on how a customer might respond to a piece of clothing.
ReCode founder Kara Swisher described herself as a StitchFix conundrum during an interview with Hudson Bay’s CEO Helena Foulkes at Shoptalk. Swisher, whose stylist finally came up with a few items she decided to keep after many fix-misses, asked Foulkes about her company’s philosophy on customer data and how it informs their AI initiatives and overall planning. While not addressing artificial intelligence directly, Foulkes described the customer/brand data exchange as making it possible “to know what you love and to make it more relevant, and easier for you to shop with us”.
Investment in AI for Retail is on the Rise
Shoptalk holds an annual session entitled “New Insights on Retail from Top Universities.” This year George Washington University Professor Donna Hoffman gave a detailed overview of both the current utilization and future integration of AI in retail. In her short but chock-a-block presentation, Hoffman shared some statistics gleaned from a 2018 report by Capgemini, which stated that four percent of retailers deployed AI in 2016, 17 percent in 2017, and 28 percent in 2018. Further, Hoffman shared that the largest retailers (over $108 billion in revenues) are the biggest investors, and forty-one percent of the top one hundred retailers currently use AI. She discussed the efficiency gains and cited numerous AI -powered activations currently utilized in retail including automated checkout, customized manufacturing, fraud detection, shopper tracking, dynamic pricing, shopper incentives, and targeted coupon strategies.
The AI Effect on Inventory is a Win for Retailers
At Shoptalk I caught up with José P. Chan, founding team member and current board member of the retail analytics platform Celect about AI utilization in retail. According to Chan, “Prior to the use of Machine Learning, a branch of AI, retailers would use historical data and gut intuition to predict future inventory outcomes. Today, they have better tools that provide crisp guardrails, which allow retailers to make better inventory decisions.” As inventory selection is better informed through utilizing Celect and other predictive analytics platforms, retailers are increasing their efficiency as merchants, prompting lamentations from bargain hunters as inventory selection becomes more efficient and there are fewer markdowns.
Switching back to South by Southwest, Kohl’s Senior Product Manager of Artificial Intelligence and Technology, Grace Burgio, gave a concrete illustration of their new inventory AI-informed best practices. She described Kohl’s previous use of “glorified Excel-based tools” to plan assortment saying, ” It is very difficult to function this way to ensure that every store gets the optimal merchandise for the customer in every location.” Kohl’s is now working at the store SKU level using AI to manage inventory and assortment “to disrupt inventory placement, and how we sell our merchandise.”
The Rise of the Intelligent Chatbots
Burgio also touched on Kohl’s use of AI in chatbots for customer-service platforms. She said, “Kohl’s is chipping away at the highest volume chat topics and blending in AI where it is appropriate. Areas such as order tracking lends itself to an AI solution. When an upset customer is calling, and they want to vent, we know that is something that we want that handled by a human being.” The key phrase form Burgio’s comments is: “when appropriate.” The development of an AI functionality that can differentiate between these two types of human/robot interactions was a common topic at SXSW. Advances in a chatbot’s ability to interpret natural language, thus determining the correct moment for a bot to human handoff is rapidly improving. This progress in language processing is moving beyond the chatbot. In fact, the consensus among many of the speakers at SXSW was that keyboards will soon be obsolete, and most of our computer interactions will be voice activated.
At South by Southwest MIT professor Aleksandra Przegalinska discussed a leap in human-bot interaction through the development of the Google Duplex, an AI trained technology for the Google Assistant. When the technology debuted in May 2018, the verbal dexterity of Google Duplex, complete with um-humms and other very human responses, was so advanced that it was met with skepticism. The company then invited individual reporters to experiment with the technology to prove its veracity, which it did.
Artificial Intelligence as a Member of the Family
Przegalinska went on to describe the very real relationships developing between her young children and Alexa. This generation of children are growing up with Amazon’s home assistant, along with those of other platforms who answer the questions they used to ask of their parents, who read them stories and are ethereal members of the household. The relationships represent a new cultural blurring of the line between humans and robots. This is where I am going to ask you to pause and let that sink in for a minute. How do we as retail strategists plan for this emerging generation of consumers who will become active in 10 to 15 years? As we sit in conferences and absorb these shifts and technological developments, we need to recognize that the reality that what seemed like science fiction in the past is fast becoming reality. The time to play out the possible effects of these changes and begin to model for them is now.