Instacart and Shopify Are Testing ChatGPT. Will It Pass the Experience Test?
By: Jenn McMillen
Will ChatGPT, the artificial intelligence chatbot, mean as much to consumers as to retailers? The tech’s ability to answer questions and then remember the conversations and context is promising in terms of personalizing shopper recommendations. But is it ready to benefit the customer experience?
(Notice: ChatGPT’s creators, OpenAI, in mid-March launched an advanced version, GPT-4, that is “more capable and accurate than the original ChatGPT,” according to The New York Times.)
Online robots can answer a lot of on-the-spot consumer questions. But can they ensure a better customer experience for retailers? ChatGPT is about to learn the truth.
ChatGPT is the artificial chatbot, launched in November 2022, that’s been getting retail attention for its ability to answer consumer questions articulately and in detail, across vast areas of knowledge. Ask ChatGPT for advice on how to decorate a living room in fantastical ways, for example, and it will produce several AI-generated images.
That’s a pretty cool party trick. But here’s the “money” part of ChatGPT: It can store the knowledge it gains while solving one problem and then apply that knowledge to a different but related problem. In short, it can recall conversations and context, and therefore can pick up with customers where they left off.
And here’s the money part for consumers: ChatGPT does all this without the influence of advertisers, according to a story in Modern Retail. It’s also free, for now, to those who register for an OpenAI account.
The ChatGPT Pile-on Is On
A major caveat regarding ChatGPT is that it is not linked to the Internet, and therefore cannot search for information online. Rather, ChatGPT generates responses to questions using information it was trained on from a large body of sources, including Wikipedia, news articles and books.
Nevertheless, retail companies, including Shopify, Instacart and Carrefour, have hitched up to the ChatGPT wagon. They’re testing videos and other features that answer questions, provide product information and make recommendations.
These are services shoppers already expect from retailers, however. The task retailers that use ChatGPT must keep in mind is: How does it produce a better customer experience?
Let’s Look At Experiential Use Cases
ChatGPT has one advantage already: Consumers speak fluent robot: 54% have daily chatbot interactions, and 49% find those conversations trustworthy – across all sectors, according to research by Capgemini.
But in retail that trust level drops to 44%, possibly due to marketing influence. This suggests that shoppers may trust ChatGPT’s advertising-free technology more than others. Here are some ways it can parlay that trust into better customer experiences.
As an always-thinking personal shopper. Because ChatGPT gets to “know” people’s preferences via questions and conversations, marketing experts believe it has the makings to become an amped-up personal shopper. Early reports indicate it has the potential, for example, to recall how a customer responded to an earlier product suggestion and build on that reaction to fine-tune future recommendations. The grocery delivery platform Instacart, for example, is testing a feature called Ask Instacart that ideally will answer grocery customer questions about product pricing, nutrition and even recipe prep time.
As an AI event planner. If retailers alert their customers to upcoming events, and customers ask ChatGPT about them (for ex: “Nordstrom offers free makeovers. What is a makeover?”), it will answer the question in detail. It also should be able to retain that information so it can suggest related activities in response to future queries. ChatGPT also can recommend hobbies, something the AARP points out to its readers, so retailers might want to promote their offerings by shopper category and age group. Note, however, that ChatGPT cannot determine the date, ZDT.net tells us. So, it can’t tell you what free events are happening “tomorrow.” It will, however, suggest resources.
As a robot dietician. An intelligent bot should be able to detect and catalog changes in a customer’s grocery lists, especially based on past “conversations.” Notable differences in product queries might reveal new medical needs (diabetes) and personal preferences (a child becomes a vegetarian). In addition to answering questions about recipes and ingredients, ChatGPT could use these changes to adjust product recommendations and inform more pinpoint suggestions, such as: “Twinkies contain animal product,” or “If you have diabetes or are watching your sugar intake, it’s important to be mindful of the sugar content in Noosa yogurt.” (That’s verbatim.)
By reducing growing pains. If a shopper asks ChatGPT about size 2T clothes today, it should have the AI to predict that person will be looking for size 3T clothing soon. However, patterns would indicate more complex needs of growing children. A young boy who stays in one size for a long time might be a late bloomer and want more “grownup” clothes, or a 7-year-old experiencing early-onset puberty might outgrow her little-girl clothes too quickly. Through style suggestions, ChatGPT can narrow preferences and use the information to find the latest trends in smaller and larger sizes (Gaps Kids is among its ideas). For parents looking for age-appropriate clothing for their fashion-focused pre-teens (thanks, TikTok), ChatGPT can list style influencers, and fashions, for that age group (while emphasizing it does not endorse them).
By serving people with disabilities. People who travel by wheelchair, who live with autism or who were born with Down Syndrome have different apparel needs for physical as well as sensory reasons (a potential aversion to tags and seams, for example). ChatGPT can identify these consumers, either through a retailer’s “tell us about you/your family” survey or through past queries and pull from its memory the brands and retailers that offer goods for such needs. For example, when asked “Where can I find clothing for a child in a wheelchair?”, ChatGPT provided several suggestions, including the Tommy Hilfiger Adaptive line and Target’s Cat & Jack adaptive clothing for kids.
ChatGPT Might Be Impressive, But It’s Not Human
Innovations, such as ChatGPT, can resolve a lot of consumer-understanding challenges for retailers and brands, but these companies should not forget: An algorithm will never know a customer as a human being. Each person it encounters is a compilation of categories – married, professional, homeowner, TikTok user, parent, diabetic.
The task remains with retailers and brands to ensure the algorithm is structured first to identify a person’s need, not to sell their own goods. The goal should be to understand the “why” behind the many searches, questions and conversations people feed the bot.
This area of smarts – emotional intelligence – is the connective tissue between an artificial and human experience.
This article originally appeared in Forbes.
Follow me on Twitter or LinkedIn.
Forbes.com retail contributor Jenn McMillen is nationally renowned as the architect of GameStop’s PowerUp Rewards, and is Founder and Chief Accelerant of Incendio, a firm that builds and fixes marketing, consumer engagement, loyalty and CRM programs. Incendio provides a nimble, flexible and technology-agnostic approach without the big-agency cost structure and is a trusted partner of some of the biggest brands in the U.S.