ChatGPT Won’t Replace Retail Customer Service Agents

Ever since ChatGPT was released near the end of 2022, organizations and decision makers have been scrambling to figure out how they can use the new generation of artificial intelligence (AI) technology to work better, faster and more effectively. Customer service delivery is no exception.

It really does feel like a breakthrough moment for technology, but retail business leaders should pause and reflect on lessons learned from earlier attempts to substitute tech for human contact center agents before moving forward too hastily.

Customer Priorities

Customer satisfaction is still the name of the game, and when it comes to customer service, retail consumers prioritize efficiency, accuracy and trust — though not always in equal measure or in any particular order. Experience has taught us that technology’s vast computational power makes it better than humans at delivering efficiency, while uniquely human capabilities like empathy, intuition and judgment make live agents better than machines at ensuring trust. As for accuracy, that’s turned out to be far harder for AI than people expected because AI has really struggled with precision.

There’s general consensus that online chatbots work fine for simple transactions like paying bills, scheduling appointments, or booking hotel rooms. In those situations efficiency is what matters most, and chatbots are better and faster than humans at sifting and evaluating data.

But when it comes to more complex transactions with more at stake, who among us would accept a rejection by a chatbot? Resolving a billing overcharge or a refund request demands accurate understanding of context and a precise reading of related circumstances. Chatbots are not (yet) very good at those things.

Efficiency vs. Trust

A recent personal experience illustrates my point. As is usually the case, I was happy to use a chatbot to book a hotel room for an upcoming family trip. This time, unfortunately, the website I used turned out to be fraudulent (not the chatbot’s fault); when my family and I arrived at midnight to check in, the hotel had no reservation for us. We were exhausted and without a room.

I called my credit card provider to cancel the fraudulent charge and began searching for another hotel room. In both cases I bypassed the chatbot option and went straight for live agents. This was no time for glitches or errors. I needed to trust that my credit card provider would accurately assess the situation and reverse the charge, and that a live agent would understand my dilemma and find my stranded family a room right away.

Can’t Have One Without the Other

Technology and human agents both contribute value to the customer service experience, but neither is capable of satisfying all customer priorities in all situations. Together though, the sum of their capabilities is greater than the parts. ChatGPT and similar technologies will alter the business landscape, including contact centers. However, before setting a course in this new technology landscape, customer service leaders should focus on what these tools really are (turbo-charged data processing prediction machines) and understand what they’re not (straight substitutes for human intelligence).

Retail industry contact centers can maximize this opportunity by steering transactional inquiries to their newly reinforced chatbots and leveraging AI’s computational muscle to bring a significant efficiency boost to live agents’ work. Customers, agents and organizations alike will benefit enormously — as long as we deploy AI to work in conjunction with, rather than in place of, human agents.

Matt McConnell is chairman and CEO of Intradiem, a provider of intelligent automation solutions for customer service teams.