ChatGPT seems to have taken over the world this year. Other forms of AI are now gaining in popularity, but they seem to be a little forgotten. But AI and automation will be more important in retail.
McKinsey found in 2017 that only 20% reported using AI to run their business. By 2022, this number had more than doubled. This year, the World Economic Forum predicted that AI services in retail would increase from $5 billion up to $31 billion.
In order to compete in the retail sector, AI has become more important. AI in retail is the use of automation, data and technologies like machine learning (ML), to provide highly personalized shopping experiences for consumers. Self-checkout technology is a significant application of AI. It offers a secure scanning system that helps to prevent shoplifting.
These AI retail platforms are able to run without the need for human assistance, giving customers greater control over their shopping experience. As an incentive for loss prevention, new systems using AI authentication will log data about suspicious shoplifters.
Modern Retail Success: A Key Ingredient
Teams that work in e-commerce marketing are also under pressure to meet the requirements of digital merchandising. AI can automate many manual tasks, saving you a lot of time.
Artificial intelligence is also expected to become a major competitive advantage. AI can provide data to support better product recommendations, omnichannel strategies and long-term business objectives.
OpenAI’s ChatGPT isn’t the only tool that retailers should consider. ChatGPT is not a substitute for other AI forms, says Eli Finkelshteyn. Constructor, an online platform that allows retailers to search for and discover products. Contrary to public opinion, ChatGPT does not completely push other forms of AI aside.
If you ask ChatGPT, you’ll get the same reply. Imagine how many times you would like to talk to a sales associate in a real store. ChatGPT is a tool that creates this back-and-forth dialog online,” Finkelshteyn explained to the E-Commerce Times.
Balance AI tools for detailed product discovery
Certain situations are when speaking to an associate via the internet is helpful and appropriate. It’s not required for all products, he said. “When you are shopping online, if you know exactly what you are looking for, search is still — and will remain — a better tool,” he argued.
A category page will still be a better option if you want to simply browse an aisle and see what shirts, electronics or snacks are available. You may want to talk to a store associate to explain your needs in more detail. This is only 5%-10% of the time.
ChatGPT is a superior ecommerce solution than what we currently have, he said. Understanding your long-form question, answering your queries, and helping you to find what you are looking for can help you close the deal.
Leveraging AI to Customize Shopping Experiences
AI can rank and customize search results based upon what it knows of trends in a product catalog. It can also tap into what the AI knows about shoppers based on their behavior on a brand’s site and other channels.
When someone searches “shirts,” the results may prioritize shirts that are in the customer’s preferred colors, brands, and price ranges. This is done while optimizing for retailer KPIs such as revenue and conversions.
Constructor, for example, uses AI and machine-learning to extract details from each user query. Finkelshteyn said that it can also incorporate the feedback provided to the system by the user after the query, via interactions with the results.
AI in E-Commerce: A Discussion
Eli Finkelshteyn was asked by the E-Commerce Times to explain what e-commerce firms can expect in terms of AI over the next few decades. He knows the traps that retailers should avoid when they consider AI.
E-Commerce Times : How can other AI options succeed in the shadow of ChatGPT?
Eli Finkelshteyn: ChatGPT may be amazing, but it’s not the right tool for every AI issue. It is not the most suitable tool to return large numbers of results or filters, such as in an e-commerce search.
It is also not the best tool to use site traffic and clickstream appropriately. [behavioral] Data to rank those results or learning how to use clickstream and user data from an ecommerce site to decide on the changes in rankings over time, or to personalize them for a specific shopper. It makes sense since ChatGPT doesn’t try to solve any of these problems.
These are the use cases in which clickstream-based AI, trained on the traffic of an ecommerce site, will continue to be much more suitable so that retailers can improve shopper experiences and create cohesive omnichannel experience.
What can retailers look forward to as AI and retail continue to advance?
Finkelshteyn: They both promise a thrilling future. ChatGPT can be combined with existing AI forms, including clickstream-based AI. This opens up new possibilities for creating new user interfaces.
For example, if you need to explain on an e-commerce website that you are looking for products for a specific purpose or event — like clothes to wear for your friend’s engagement party or ingredients to make a casserole — that is currently difficult. There are currently no good methods of finding products, like browsing and searching category pages.
Innovations can be brought about by incorporating ChatGPT, which helps retailers better understand the long-form questions and abstract requests of shoppers. The idea of using clickstream-based AI for product discovery to provide trending personalized results that meet the user’s requirements is exciting.
What impact has the popularity of ChatGPT or its technological advancements had on Constructor’s product search and discovery platform?
Finkelshteyn: In general, people are still trying to understand the place of ChatGPT within ecommerce. Experimentation and flexibility are required to find the best interfaces and use cases. It is important to make sure that the ChatGPT technologies you offer to your users are truly helpful.
What does Constructor do to promote such innovations?
Finkelshteyn: We provide an AI-based product discovery platform to help retailers deliver highly customized experiences across all channels, which also reflect their KPIs. Constructor’s work has been impacted by ChatGPT in that our retail partners have expressed an interest in trying our prototypes and are willingly committing development resources faster to integrate them than previous new technologies.
For years, we have believed that product discovery could be improved over what customers are used to. ChatGPT encourages retailers to experiment with different forms of product discovery, and see which ones their customers like best.
How can retailers ensure that the real benefit to shoppers is not lost in all this experimentation?
Finkelshteyn: Unfortunately, many of the efforts we see so far in the space — integrating ChatGPT with other retail technologies — have been gimmicks. These gadgets may look cool when you first play with them, but they’re useless for shoppers.
A company launched a chatbot using ChatGPT. It basically asks users to search for a product and then prompts them with their needs. It’s actually less useful than using the search bar. It doesn’t provide any real value. Shoppers will try it a couple of times and then abandon it.
ChatGPT’s energy and its impact on the product discovery industry give us all the opportunity to revolutionize our space with new interfaces, technologies, and tools. Constructor’s goal is to harness that energy to develop technology that shoppers will use repeatedly, and not waste it on gimmicks they will only use once.
What can e-commerce businesses expect from AI over the next few decades?
Finkelshteyn: Expect to see a lot more experimentation. ChatGPT and, generally speaking, the technologies behind it — transformers and large language models (LLMs) — make a lot possible that was not possible previously. Just because something is feasible does not mean that it will be valuable to an ecommerce company or their customers.
Experimentation will be necessary to discover the most effective places to implement this technology. Constructor is already utilizing it to enhance the context of the results that are returned to users. This is just the beginning.
What should retailers keep in mind when they implement AI tools and what should be avoided?
Finkelshteyn: Before anything else, ensure that what you do has the potential of being valuable to both you and your customers. ChatGPT is already full of gimmicks, but there will be many more.
Before you make a purchase, carefully consider the benefits that ChatGPT or other AI-based technology can provide. Consider the pain points and problems you want to solve.
You might consider connecting data across channels, reducing labor-intensive tasks in merchandising or decreasing search abandonment. Consider your best use cases for AI.
Where is ChatGPT most useful, and clickstream AI a better choice? What other tools, such as an intuitive tool for merchandisers and business users that are not AI-based, might be better suited?
This space will see the arrival of valuable tech. You can also expect to see many more so-called “innovations”, which may look cool, but will not be helpful or worthwhile. Retailers who are able to spot the differences will be most successful.
What are your concerns about the safety and ethical implications of AI increasing use? Several European countries have recently banned the use of ChatGPT.
Finkelshteyn: Conversational AI is one form of artificial intelligence that can deceive and lie confidently. Deepfakes and generative AI are already making it more difficult for people to trust their ears and eyes.
Should governments be more active in monitoring AI technology
Finkelshteyn: The government cannot and should never rely on AI firms to legislate for them. Instead, they must do everything they can to educate and protect their citizens. It is crucial that those in power in companies on the cutting edge of AI use AI ethically. It is reasonable that governments ensure this happens.
What impact could government intervention have on the future development AI?
Finkelshteyn: This is a double-edged sword. We live in an increasingly globalized world, where internet access is available to billions across nearly every country. This makes AI hard to police. In the short run, a government that slows down AI research conducted by companies within its jurisdiction may protect its citizens. But in the long term, it will hurt them.
The rapid advancements in AI will give companies in countries who fail to take the same decisions a huge advantage. We are still all connected to the Internet. AI developed by less scrupulous countries will still impact people all over the world.
I hope that our leaders will rise to the challenge of finding the right balance. For those of us who lead AI companies, it is important to constantly examine what we are working on and ensure that it’s ethical. We should also encourage our peers do the same.
Editor’s note: To learn more about the various issues surrounding Ai development and regulation, please see The AI Revolution Is at a Tipping Point.