Shah used the software to improve his in-house customer service chatbot Lina, training it largely on his company’s help center content. In December 2022, he let the ChatGPT-powered bot field nearly all messages and found his customers were largely happy.
By June, he had fired 27 of his customer service agents and replaced them with Lina.
“It was [a] no-brainer for me to replace the entire team with a bot,” he said in an interview, “which is like 100 times smarter, who is instant, and who cost me like 100th of what I used to pay to the support team.”
ChatGPT took their jobs. Now they walk dogs and fix air conditioners.
Call centers, which started becoming popular in the late 1990s, provide cheap offshore labor to handle mundane corporate tasks for Western companies. Now, amid a larger wave of automation, this work is being outsourced once again: to artificial intelligence.
Backed by advanced technology, chatbots can interact directly with customers and solve problems independently — no human necessary.
Start-ups, banks and consumer goods corporations say the technology can help them cut costly customer service costs while providing clients with personalized service. They argue AI will give remaining call center workers more support, helping them feel fulfilled in their jobs.
But economists and workforce development experts say the shift could have a profound effect on economies across the world, especially in countries like India and the Philippines, where call centers provide millions of people with modest-paying work and where surveys have shown automation could render over a million jobs obsolete. The change is sparking a debate in the Global South about what, if anything, they can do to prevent a mass workforce disruption.
“You will … end up seeing a lot of jobs be eviscerated,” said Sharad Sharma, co-founder of the iSPIRT Foundation, an Indian nonprofit technology think tank. “And the fewer jobs that will remain will be different kinds of jobs.”
Outsourcing gained speed in the 1990s as global economies became more open and computers made it easier for work to be done anywhere. American companies, looking to control costs, began shipping jobs that were necessary — but not core — abroad, where cheap labor was in ample supply.
Countries such as India, with well-educated, English-speaking young workers, became a popular destination for these back-end jobs — most notably in customer service, information technology and data-processing roles.
The economic impacts were significant. In the early 2000s, Western companies employed just a few thousand people in the Philippines, but as of 2023 that number has ballooned to over 1.6 million, according to industry analyst figures. In India, outsourcing jobs now contributes to nearly 10 percent of the country’s gross domestic product, industry analysts said.
The emergence of generative artificial intelligence — technology that undergirds AI chatbots, image makers and voice-cloning software — threatens the industry. Customer issues are likely to be solved by software that can analyze millions of data points to find the most common solution to a problem, rather than human ingenuity.
In a September survey of more than 300 IT and call center leaders, 80 percent said conversational AI products will be a “must have” for their companies going forward, with 52 percent of polled firms reporting they had already invested in the technology.
See why AI like ChatGPT has gotten so good, so fast
The shift has countries who rely on call center work bracing for change, politicians, economists and workforce development experts said.
In May, Imee Marcos, a senator in the Philippines called for an inquiry into the potential displacement of workers by technology, especially in the country’s outsourcing sector. She noted an Oxford Economics and Cisco study, which estimated that digital automation could render 1.1 million jobs in her country obsolete by 2028.
“AI is developing faster than most people can comprehend,” she said in a news release, “and is threatening to take away jobs and turn employment growth upside down.”
In India, companies are already testing out how to replace customer service workers and call center agents with large language models, said Sharma, though he declined to disclose which companies he was aware of that were doing these kinds of tests.
He noted that with agents using AI to augment their work, the amount of workers needed to handle the same volume of calls will dwindle.
The effect to the economy will be undeniable, he said, and Indian outsourcing companies will need to adapt artificial intelligence into their operations quickly if they seek to keep jobs from going elsewhere. “India has to embrace augmentation faster than any other country,” Sharma said. “That is the conversation that is taking place.”
But AI and labor rights experts are torn on the impacts.
Erik Brynjolfsson, a professor at the Stanford Institute for Human-Centered AI, said automation could be a boon to the industry. As part of his research on AI’s impact on call centers, he found that many companies are augmenting their agents’ jobs with AI software instead of fully automating them.
Brynjolfsson said a suite of AI software tools help call center workers do their jobs better: transcription software analyzes call conversations in real-time and provides agents with ideal solutions; other tools identify angry customers and provides techniques to calm them down.
This support allows workers to tackle difficult problems with far more ease, he said, leading to higher customer satisfaction and lower worker turnover. In his study surveying more than 5,000 call center agents using artificial intelligence software, access to the tools increased productivity by 14 percent, with novice workers seeing the largest gains and highly experienced workers finding negligible improvement.
He said companies should take note of these findings. “Don’t look first to automate,” he added. “Think about where you can augment things.”
ChatGPT ‘hallucinates.’ Some researchers worry it isn’t fixable.
Virginia L. Doellgast, a labor relations professor at Cornell University, argued the technology will make jobs more difficult: Easy customer service problems will be handled by artificial intelligence chatbots, leaving more complicated issues for humans to deal with.
Remaining workers will deal with more difficult tasks that take longer to solve. If corporate compensation structures are tied to metrics, such as shorter call volume, they could be financially penalized.
If artificial intelligence makes it easier for novice workers to perform better, she said, that could also justify companies hiring less expensive, less experienced talent, leading to wage deflation across the industry and less satisfying customer service.
“Why would you need to hire a really experienced center worker who costs more money for their skills when you could get the same performance out of a less experienced one,” she said.
But for entrepreneurs like Shah, cost matters above all. After the success of his Dukaan chatbot, he created a spinoff product that allows others to transition to an all-AI-powered hotline.
Not every customer service employee should worry about being replaced, but those who simply copy and paste responses are no longer safe, according to Shah.
“That job is gone,” he said. “100 percent.”