Study: Generative AI boosts productivity, retention

A study of a generative AI system used by 5,000 customer support agents found that those workers experienced an increase in productivity and retention, and a reduction in stress. But the workers who gained the most from this automation were newer and less experienced. 

Customer service work is aptly suited for generative AI, according to this just-released study from the National Bureau of Economic Research, a nonprofit research group based in Cambridge, Mass. 

Customer support is a stressful job. "A key part of agents' jobs is to absorb customer frustrations while restraining one's own emotional reaction," the paper noted. But generative AI can act as an aide, using the customer's chats as input and providing suggestions for empathetic responses and problem solving in real time. The study found that the use of generative AI reduced the likelihood of customers wanting to escalate issues to a supervisor. 

The research paper's authors are Erik Brynjolfsson, an economist and director of the Stanford University Digital Economy Lab; Danielle Li, an associate professor at the MIT Sloan School of Management; and Lindsey Raymond, a Ph.D. candidate at MIT Sloan. 

The company that allowed researchers to study the effects of generative AI on its contact center wasn't identified, other than as a Fortune 500 that provides business process software. Customer service work generally includes everything from product support to answering employee HR questions. 

It's too soon to start making conclusions about where this will have an impact and how big that impact will be.

The increase in productivity -- nearly 14% -- wasn't surprising to industry analysts, but it also has limits. 

David Creelman, CEO of Creelman Research in Toronto, cautioned against putting too much weight on one study. "It's too soon to start making conclusions about where this will have an impact and how big that impact will be," he said.

"But it does seem reasonable to expect that it'll be easier to get these new technologies to do an average job than it will be to get them to do an excellent job," he added. "This means that new or lower-performing employees who are doing an average or below-average job are more at risk than those who are doing an excellent job."

He compared generative AI to a grammar checker, which helps someone avoid errors, "but it won't make you a great writer. "

The study also stated, "the AI tool we study is meant to augment -- rather than replace -- human agents." 

The U.S. Bureau of Labor Statistics expects customer service representatives to decline by 4% through 2031 because of automation. 

Ian Jacobs, a research director at Forrester Research, said the consultancy expects generative AI to increase customer team productivity. But the technology is "not yet automating away most human-to-human interactions," he said, pointing to its accuracy problem. 

"The biggest boon for productivity from generative AI today comes from augmenting the humans who actually serve customers," Jacobs said. "Agent augmentation is the name of the game right now."

He said generative AI can summarize a body of text effectively, saving agents time and letting them move on to their next calls. 

Ben Eubanks, chief research officer at Lighthouse Research & Advisory, said he believes generative AI "will make HR better and faster, just like chatbots have done."

Generative AI plus humans can help with tier-one support requests, which are routine, common and straightforward. He said the more complex tier-two requests, which are specialized, unique and unexpected, will need to be addressed by HR professionals. 

Patrick Thibodeau covers HCM and ERP technologies for TechTarget Editorial. He's worked for more than two decades as an enterprise IT reporter.