Photo: iStock / Vertigo3d
Responsible And Practical Ideas On How To Apply AI And ML
Artificial intelligence (AI) was at the top of the heap in terms of buzz on the Shoptalk showroom floor and topics and questions in the sessions at this year’s conference in Las Vegas. A recurring joke at the conference was that some words become so commonly mentioned that it becomes like buzzword bingo. Unlike past terms/technology that fit this description, AI seems to have legitimate staying power and use.
A session on Monday, “New Applications of AI and Machine Learning,” featuring Patrick Duroseau from Under Armour, Barkha Saxena from Poshmark and Anshuman Taneja from Peapod Digital Labs, was moderated by Lauren Wiener from Boston Consulting Group in a lighthearted but informative way.
There are mixed perspectives on AI in the market now. Some think it will be revolutionary and others are cautious, worrying that it will take over the world. Mr. Duroseau provided a balanced approach, saying AI is just like any other technology: “You just have to understand the context where it belongs, how to absorb it, how to leverage it and really just embrace it the proper way.”
Two key elements of embracing it the proper way are collaborations with other leaders and promoting and executing in a responsible manner. Companies need to work together to forge consensus on the appropriate governance and usage of data and algorithms with a technology that is quickly evolving. Mr. Duroseau said it is important for companies to participate in organizations that unite to promote the responsible use of AI, such as the Data & Trust Alliance.
Ms. Saxena continued the theme of establishing and managing governance to ensure good use of human-level AI models. She explained that it is hard to answer all questions using AI for chatbots. Simple answers are made for AI chat. Individual associates with specific topic expertise, however, are best at responding to specific second-level questions.
Mr. Teneja said there are opportunities for retailers to deploy AI to improve customer experiences and supply chain performance. Grocery has some of the highest frequency of visits and repeat purchases, which drive a lot of data about types of trips and varieties of baskets that can continuously feed machine learning efforts. There is also a great deal of data on promotions, pricing, loyalty, digital touchpoints and other influences on shopping. This volume and completeness of data can go a long way in using AI for convenience.
Discussion Questions
DISCUSSION QUESTIONS: Are there retail verticals – apparel, grocery, et al. – where artificial intelligence will have a more substantial impact? How can retailers determine the best use of AI and human resources to optimize results?
Poll
Loading …BrainTrust
Matthew Pavich
Sr. Director Retail Innovation at Revionics, an Aptos Company
David Naumann
Marketing Strategy Lead - Retail, Travel & Distribution, Verizon
There are numerous ways that AI is having a meaningful impact on retail right now. Gartner indicates that AI-based pricing is the top use case for AI in retail and grocers, drug stores and convenience stores – where data is very rich, as the article states – have taken advantage of the benefits of AI for years. Generative AI, however, is much newer, and likely the source of much of the buzz at ShopTalk. And with generative AI, I think site search is the first and most significant way to derive meaningful value. More on my opinion here.
As Dave has stated, AI is a powerful tool that is driving meaningful value today across multiple retail verticals. Although there is a lot of hype around generative AI and some of the more creative and qualitative aspects of AI, there is no doubt that the quantitative aspects of AI are the most important ones for retailers today. In fact, Gartner specifically calls out 26 different use cases for AI in retail with price optimization, promo optimization and demand forecasting as the clear top use cases for value and feasibility. In a highly inflationary market with shifting brand loyalties and endless disruptions, the data is clear — retailers using sophisticated AI solutions for their pricing have stolen share and customers profitably from retailers who haven’t made the leap. While a lot of AI capabilities may not be a sound investment today, there is no doubt that retailers trying to compete on price without sophisticated tools are at a strategic disadvantage.
Historically, applications of revolutionary technology are first applied to increase the efficiency of existing processes. It is only after the familiar and obvious uses are accepted and routine that novel applications typically create new business processes, industries, and shareholder value. Initially, copywriting, automated custom-tailored email marketing campaigns, etc. seem to be the current low-hanging fruit, so I’d anticipate that larger retailers with D2C and/or curated operations would be good candidates. These companies would do well to create off-site “skunk works” groups of bright, tech-savvy department specialists and let them go wild.
As mentioned in the article, retail segments with frequent purchases and high volumes of data are prime candidates for AI. AI for chatbots, in my opinion, has not been very successful, as I don’t recall any instance where a chatbot answered my question correctly. AI is a great tool for analyzing massive amounts of data to improve predictive analytics for forecasting and offering personalized recommendations or offers based on historical data.
As the article suggests, “Two key elements of embracing it the proper way are collaborations with other leaders and promoting and executing in a responsible manner.” I fear that irresponsibility will cause a slowdown of this remarkable technology and nefarious actors will put limits on its near-term usefulness.
With that said, any retail operation is a prime candidate for its application (e.g., customer service, merchandising curation, and theft control). I can’t wait to see how this technology evolves and delivers heretofore unrealized advantages.
Elon Musk and Steve Wozniak and over 1,000 experts have said AI can “pose profound risks to society and humanity.” For months now, AI has been front-page news. Have we watched too much sci-fi? Ex-Machina, yikes? Or do these futurists understand what can develop shortly versus most of us who could not anticipate this conversation just a few years ago? Most of us had no idea of the depth of AI and ML being used in manufacturing.
Retail is leagues behind traditional manufacturing in terms of implementing AI and ML into operations. At some point, if they do, every aspect of retail will be affected and become more dynamic. Everything from getting products to stores to having the right offering for the customers. Can it be done? Yes! Will it be done? That is an entirely different discussion.
AI is a prime example of “what’s old is new again,” and how much of what was talked about at Shoptalk were ideas that have been mentioned in the past, but not utilized correctly. I was happy to attend this session and see that the AI applications retailers were discussing are really strategic.
Data and outcomes are critical factors for AI/ML and responsible AI is an extraordinary step that most data and analytics companies are ignoring at their peril. Belonging to organizations that think about ethical factors won’t be enough- the company using the AI needs to build it into their own engagement. AI is use case driven and hence can’t be applied to retail sub-verticals uniformly nor can a sub-vertical define the usage levels. AI will continue to have substantive impact, but in most cases it will be embedded into software and invisible to end users. Retailers will need the support of AI engineers and analysts to decipher the complexity of AI, but it will also need folks who understand data, ethics and implications to decrypt the human element.
AI is already having a substantial impact on the retail industry across the board, regardless of vertical, because the best use of AI is to perform tasks at a scale that humans simply cannot keep up with. And many of these tasks, like search tuning, are common across all kinds of retailers – both B2C and B2B.
Most large retailers have thousands of product listings on their e-commerce sites and a high volume of search queries each day. It’s simply not possible for humans to keep up with the amount of data generated. A merchandising team may only be able to optimize the top 5 percent of searches, leaving 95 percent of the site unoptimized. AI, however, can optimize the entire site, leaving the merchandising team free to focus on more strategic business objectives, while also driving sales, revenue and conversions. So the most substantial impact of AI is twofold: processing vast amounts of data at scale in ways humans can’t, removing lower-value tasks so their teams can focus on the higher-value initiatives, while also optimizing that data at scale for key e-commerce metrics.