How can AI help you? What LLMs mean for the future of customer service jobs

Ask any executive where they’d most like to put a chatty AI to work, and there’s a good chance they’ll say customer service.

The versatility of chatbots powered by large language models (LLMs) can seem like a natural fit for fielding gripes and headaches, which could spell trouble for call-center or other support workers. So will the next generation of AI assure that you never talk to a human representative again and that headset workers will be out of a job?

Experts say it’s not so simple. One thing that’s certain is that customer service jobs are here for the foreseeable future—and the field may always require some form of human touch. But AI is already starting to change how some of these workers do their jobs in the form of generated scripts and call summaries.

Research firm Gartner predicts that generative AI will prompt a 20%–30% decrease in the number of customer service and support agents by 2026. Customer service was classified by Pew as one of the most likely jobs to have “medium exposure” to AI, meaning the technology could replace workers in the sector or enhance their productivity. And McKinsey found that the field was one of those most at risk for automation-related job losses.

“[Generative AI] is just really going to reshape customer service in ways that we don’t even understand yet,” Forrester Principal Industry Analyst Max Ball told Tech Brew.

A “fear dam”

All of these sweeping predictions come as companies like Expedia, AT&T, and Delta Air Lines talk a big game about deploying fine-tuned chatbots or other AI-powered customer experience tools tailored to the specific needs of their user bases.

But customer service workers may be able to rest easy, at least for the time being; Ball said many of these projects are still “lagging” where execs might have imagined they’d be by now. There’s a “fear dam,” wherein businesses are still too concerned about risks like hallucinations to fully automate their operations.

“The brands I talk to, the conversations there, up until July—I make this joke, and I never got this actual request—but they were all along these lines: ‘Somebody in the C-suite said, Hey, let’s get three ChatGPTs and fire all our agents. Can we do that by Tuesday?’” Ball said. “And now it’s more, ‘OK, I kind of understand what this is, where’s it going to fit? How’s it gonna work?’ It’s early.”

The appetite for this tech is clearly there; however, Ball said customer service is seen as a ripe area for cuts among companies aiming to trim spending, and once they can feel confident that AI is reliable enough to handle these tasks, the equation will likely change.

“The interest in self-service is way up there because the cost savings are so huge,” Ball said. “As soon as the fear dam gets broken, it’s gonna go crazy.”

Press 1 for a human touch

Not everyone agrees on what exactly the jobs impact will be if this happens. Cory Stahle, an economist at jobs site Indeed, said his recent research on AI and work looked at fields like customer service differently than some of the other reports out there.

Stahle said his team aimed to dig deeper into the skills commonly listed for customer service jobs on Indeed and evaluate how much exposure each attribute might have to AI.

“When we look at customer service, what we see is that, yes, generative AI has the potential to really impact those jobs to potentially help aid those workers by offering things like chatbots,” Stahle told Tech Brew. “But many of those human skills that are involved for customer service workers are still going to be required, even if chatbots and some of those things become a more normalized part of these roles.”

Researchers at Stanford and MIT studied the effects of a fine-tuned chatbot deployed in the call center of a Fortune 500 software company, and their paper earlier this year found that the technology could boost agent productivity by an average of 14%. But they also found that the technology had a “minimal impact” on the productivity of highly skilled and more experienced workers.

“Indeed, we find evidence that AI assistance may decrease the quality of conversations by the most skilled agents,” the authors wrote.

An uncertain future

A caveat offered by everyone we talked to is that it is still very early days for the technology, and it’s hard to know just how fast it will develop. Daniel Colson, co-founder and executive director of the AI Policy Institute think tank, said call-center workers are likely to be more exposed if the technology is able to get better at mimicking human traits.

“I think [that] once you get the personality interface worked out for these chatbots, I think that’s going to really change the game to the customer service stuff,” Colson said.

With millions of people employed in call centers or other customer service–related professions, any job losses could have a big impact. While Colson said that, historically, the public has “generally been willing to tolerate” job losses around technology when the benefits are apparent, a significant percentage of workers being displaced by AI could create “a real political hurricane.”