Adopting ChatGPT and Generative AI in Retail Customer Service
By Robin Gomez
Director, Customer Care Innovation, Radial
In the last few months, ChatGPT has fascinated the world with its promising potential and impressive capabilities, but what does this mean for retail today? Should you adopt it now or is there more to consider given how new and somewhat controversial this form of AI is?
What is ChatGPT? Chat GPT-4 is a form of generative artificial intelligence (generative AI) developed by OpenAI that can create content based on large language models — a process that trains artificial intelligence on large datasets prior to it being deployed by users. Because it is pre-trained, it is able to be put to use out of the box with little to no training or tuning by data scientists. This makes it super easy to deploy for retail businesses as they look for ways to leverage GPT’s potential. Generative AI is not new. However, one of the biggest differences between previous versions and ChatGPT, is that ChatGPT has much greater conversational AI capabilities due to its pre-training. It’s an exciting time in AI and certainly, a revolutionary one in retail.
As the director of Customer Care solutions for Radial customers, I interact with many eCommerce retailers that want to jump onboard the generative AI tool bandwagon and start leveraging its potential. I’m having conversations with customers about this unexplored territory and what the journey looks like.
We all have questions, but there seems to be a consensus that this evolution in AI technology is too big to ignore. Whether it’s just having a position on ChatGPT or starting to implement it in your automation strategy, it seems imperative that retailers give space to this emerging transformation.
One of the questions I get is, where should we start with ChatGPT? Customer care is one of the best places to begin. We’re using it at Radial with optimistic caution and expectant hope that it will truly enable customer service to leapfrog by delivering conversational AI functionality that, so far, has been a rather pain-staking process to model, train, deploy, and tune.
That said, there are numerous applications for ChatGPT in retail overall.
ChatGPT Retail Use Cases
Some of the early-adopter use cases include:
- Personalized recommendations and upselling opportunities based on browsing and purchase history
- Autogenerated product descriptions
- Emails, text messages, notifications and marketing content creation
- Customer service virtual agent conversations and agent assist
- AI chatbots for customer order assistance
In customer support, we’re seeing ChatGPT uplevel our chatbots to make it easier for customers to communicate with self-service and improve our ability to support agents through AI systems that work alongside them. ChatGPT enables chatbots, intelligent virtual agents (IVAs), and Agent Assists to be more approachable and flexible in how humans communicate with them. Previously, our AI capabilities have confined us to using certain keywords, or request that customers use certain words or phrases to navigate self-service. This can be restrictive and has a potential for a negative customer experience if the customer cannot suggest other needs or requests outside the scope of the AI programming. ChatGPT is able to be far more conversational and comprehend a more natural, nuanced language flow which lends itself to a more human-like experience.
With its writing abilities, ChatGPT has the potential to autogenerate call transcriptions and after call work summaries for agents more accurately than current AI programming. It can help ensure that customer communication is tracked in real-time and provided to agents so they know the full story when a customer communicates with them.
These are the two major use cases we’re focused on at Radial and what I see others across the retail industry starting with as well.
That said, as enthralling and promising as ChatGPT advancements seem to be — and as much as we need to be on the forefront of a major change in technology to keep pace with customer expectations — I advise peers to be wise stewards of this new capability. What do you need to know before you strategically deploy this technology?
Before You Dive into ChatGPT…
Since ChatGPT debuted for public use in November 2022, we’ve all been a bit fascinated with it. As with any industry-disruptive technological breakthrough, it’s prudent for users to take a step back from the hype and be mindful in how they approach adopting it. There are issues with ChatGPT that retailers need to be aware of as they build their AI strategy. Personally, I recommend working with vendors with experience in generative AI models that are tailoring ChatGPT for retail use; I also believe as leaders, we have an ethical responsibility to know what we’re adopting and its potential risks.
ChatGPT is trained on publicly available internet data, which means it is not always accurate. It also has had issues with biases, and generating toxic or offensive responses, including racist and antisemitic remarks. There’s the unresolved issue around plagiarism and how to be sure content is legal to use. ChatGPT was also trained on a fair amount of material that was copyrighted and the legal implications and case for fair use have not been resolved yet. So, there are questions and currently unknowable risks in launching ChatGPT without having the proper oversight in data science terms, and in legal, moral, and ethical concerns as it relates to your brand.
At Radial, we’re working with a leading ChatGPT developer partner to mitigate risks and see to it that our usage of generative AI is appropriate, respectful, and confined to the applications we intend it for. We use best-of-breed technology partners for all aspects of our contact center platform to ensure we deliver exceptional customer experiences. Our adoption of AI aligns with our dedication to our customers.
What are some of the aspects that should be evaluated when deploying ChatGPT?
- Customer experience matters most. It’s easy for those of us in customer care to get excited about new tools and how they promise to improve call deflection, lower costs, and make life better for our agents. We assume they’ll make life better for customers, too. But keep in mind that anytime we’re deploying new technology, we run the risk of creating an (unintentional) negative customer experience. We’ve seen this with IVAs. What is meant to improve CX often frustrates users, especially when the technology is new and has glitches. You can offset this by being transparent with customers about the early adoption phase and you can invite them to provide feedback. Be sure to give customers the option to reach a human or you risk losing their business by a negative CX. Before you deploy ChatGPT, mindfully walk through the journey customers will have with it.
- Data foundations first. While ChatGPT doesn’t require data scientists to train it, it does require a solid data foundation to deploy it. ChatGPT comes with a broad knowledge base, but you will still need to train it specifically to your brand and products. It uses machine learning based on the data you provide it, which means if your data quality is bad, inaccurate, or outdated, ChatGPT responses will be, too. ChatGPT requires increased capacity for real-time data processing, data warehousing, data integration, and improved data quality — all things your brand should have anyway, but some brands will need to invest in improving their data layer foundation before they’ll see solid ChatGPT results.
- Ethics and liability. Humanity has to make collective decisions about how we’re going to use AI and what the guardrails will be. Major tech leaders have requested a moratorium on ChatGPT development until we address the ethical issues it’s raising. While your brand may not be involved in the larger arena (perhaps it should be), as a brand you do need to address these issues and be able to explain and describe your positioning to customers before you let OpenAI’s ChatGPT loose on your CX. Talk to your legal team.
- Humans and machines. AI is not designed to replace humans; but to work alongside them and alleviate them of rote, transactional work so humans can focus on higher value contributions. That said, AI will ultimately replace some human jobs, just as the industrial revolution did. But, we need to be mindful about addressing employee fears about job security, and educating and coaching them on how to collaborate and partner with AI-based workforces. I am mindful that in our pursuit of more conversational AI via ChatGPT, what we’re really after is the capacity to mimic a human to human interaction. Machines need to accommodate humans, not vice versa. And delivering better CX via machines means we need to understand the true value of human interactions, since it’s these interactions we’re trying to replicate.
- Creating an intentional place for ChatGPT. Generative AI has a ton of potential and with pop culture adopting it for a myriad of uses, it will eventually become part of everyday life. We are the guardians of how this powerful technology finds its place in our world, and for that, I believe we need to create an intentional place for it that has boundaries, defined use cases, and appropriate restrictions. ChatGPT will transform retail customer care, but we need to make sure it’s done with a strategy that truly is bound by its ability to improve the lives of customers and employees.
Get started with Radial Customer Care
We are in the early days of what generative AI can do. However, with the right boundaries in place, ChatGPT has the ability today to improve customer experience in retail contact centers. Radial provides fully outsourced customer care for leading eCommerce retailers. We are adopting ChatGPT to improve self-service and support our agents in better supporting your customers.
Please note: Radial no longer offers a Customer Care solution as of September 15, 2024.
Robin Gomez has been in the industry for 25 years, beginning with catalog-based fulfillment and customer service through the growth and evolution of eCommerce. He has worked in Human Resources, Training, Quality, Operations, Consumer Insight, Project Management, BI, CI and is currently the Director of Customer Care Innovation for Radial. He is focused on engaging and partnering with leading and emerging technology providers and curating a robust solution stack for Radial’s eCommerce focused clients and internal agents and users. Robin has a B.S. in Psychology with an M.S. in Industrial/Organizational Psychology. You can follow Robin on LinkedIn.
Looking to partner with a specialist in retail customer care?
