“Chatbots,” before ChatGPT revolutionized the world of AI, was a bit of a dirty word. To many consumers, a chatbot was a small box in the corner of screen, where a cheery automated program would offer to provide help–but then struggle to understand queries and deliver the right information.
A November YouGov survey reported that 60% of consumers felt at least fairly confident in their ability to tell a human customer service agent from a robot. And over 80% of customers are willing to wait for some period of time—for some, as long as 11 minutes—to talk to a real person, even if an AI chatbot is available immediately, according to data from Callvu, a customer service platform provider.
But now, newer AI programs are better at understanding what customers need, searching for the right information, and displaying it in a legible way. During a July 31 breakout session at Coins2Day Brainstorm AI Singapore, sponsored by Accenture, speakers shared some examples of how new AI programs could revitalize customer service. (Accenture is a founding partner of Brainstorm AI).
Generative AI programs can deliver better answers than official customer service chatbots, Joon-Seong Lee, senior managing director at Accenture’s Center for Advanced AI, claimed. Lee said that Google’s Gemini AI program helped him figure out how to navigate a bank’s system to link one account to another; the bank’s chatbot failed to understand the question.
Lee argued that websites needed to move away from a search model, where users have to go digging for answers themselves. “You’re not searching for answers. You want the answer,” he said.
Sami Mahmal, data lead for Zurich Insurance, pointed to an instance in Indonesia where the firm used AI to save time for the customer.
Indonesian law requires insurers to inspect cars before they can sell an insurance policy to the owner. These inspections are usually done in-person, meaning an owner has to wait before an assessor becomes available.
“Can you imagine? You just bought your car. It’s second-hand. You have to wait one week before Zurich comes to your place,” Mahmal said, noting that the wait extended to two weeks in some locations.
Now, Zurich asks customers to submit photos of the cars themselves. An automated process can now assess the damage and either approve a policy or refer it to an assessor for further assessment.
“We switched from a process where we had to wait days and have a manual assessment, to something that’s happening in a couple of minutes,” said Mahmal.
Will companies get a return from investing in AI chatbots?
Brainstorm AI attendees were interested in what sort of return they’d get from investing in expensive generative AI programs to improve their customer service.
While over 90% of chief information officers knew they had to make a decision on whether to use AI, more than half of them had no idea what that decision should be, noted Sinisa Nikolic, director of high performance computing and AI at Lenovo Asia Pacific.
That means Lenovo’s consultants have to help clients figure out how to help them make that decision. “What is it you want to achieve? Is it efficiency? Is it less downtime on the manufacturing floor? Is it an increase in NPS scores for client satisfaction? What is it that you want to do?” Nikolic said.
Nikolic shared Lenovo’s own experience, noting that AI had increased efficiency in its supply chain by over 80%.
Mahmal suggested that using “proactive chatbots”—programs that listen to a call and pull up important information for human agents without them needing to search for it—could reduce operational costs by between 30%-50%, and reduce call times from 15 to ten minutes.
Lee offered a different approach, noting that generative AI could improve a company’s ability to reach out to customers.
“In the past, [digital marketing companies] have run only 400 to 500 campaigns a month,” he said. Thanks to generative AI and hyper personalization, “they can do thousands of campaigns.”
