AI in customer service: How work will change

For Kayleigh Harris, who works in customer service for a credit card company, her day at work now looks very different.

Her time was occupied by making phone calls, taking notes, and lengthy card disclosures. Now, thanks to artificial intelligence, they can focus more on each customer, Harris told CNBC.

For Shannon, too, AI changed things, but not in a good way. She works in customer service for a telecommunications company in Cardiff, Wales, but did not want to give her last name for fear of being identified by her employer. She told CNBC that AI-generated notes at her current workplace often have a lot of mistakes, “so I just write my own notes.”

At her previous employer, a real estate services company, senior management instructed staff to “talk to me on ChatGPT” if they didn’t know the answer. An AI agent would book an appointment, but if it couldn’t provide the customer with the date they wanted, it would hand the case over to a human.

Staff “were constantly fighting the fire,” Shannon said. “All the complications came our way and caused a lot of burnout.”

In theory, customer service automation means that human agents will spend less time managing and more time assisting customers with issues that require human judgment, empathy, and experience.

Nikola Mrkusic, co-founder and CEO of PolyAI, a London-based company that develops voice AI systems for customer service, said contact center agents could face the “maximum abuse of the minimum wage.”

He told CNBC that while operating under strict performance metrics, they often deal with dissatisfied customers, sometimes without the authority or information to resolve the issue. He sees AI agents as a way to help shoulder that burden.

“It’s a really, really difficult job,” he said. “With AI, I think we have an opportunity to bridge the gap to make things better for both[customers and employees].”

British communications and media company Virgin Media O2 launched an AI voice agent in September that the company said was designed to do just that. Agents handle routine issues, allowing human advisors to spend more time on complex or sensitive cases.

Debra Maxwell, CEO of UK-based customer experience company ArvatoConnect, told CNBC that her company aims to have 70% of interactions handled by technology and 30% by humans.

One of the AI ​​tools the company developed for Crown Commercial Services, the UK government’s procurement agency, generates draft supplier viability assessments with 97% accuracy, reducing the need for hours of manual data extraction, she said.

As AI takes over mundane tasks, human agents may be left to handle more complex cases and be asked to pick up the slack when technology fails.

Pressure on customer service staff is increasing as workloads across the industry are reported to be increasing.

According to MaxContact’s 2025/26 UK Contact Center KPI Benchmark Insights Report, more than half of contact center decision makers surveyed said agent workloads have increased from around 40% in 2024, although the study did not reveal what was causing the increase.

Amanda Jones, a researcher in organizational behavior and the future of work at King’s College London, specializes in job intensification, the process by which job demands become more demanding over time. He told CNBC that AI could further intensify call center operations, as human agents are increasingly tasked with handling the most difficult interactions.

“Some customers may already be frustrated by the time they get to that human,” she said. “One thing we sometimes overlook is that relatively simple tasks can provide a psychological break between more demanding tasks. When these tasks are eliminated, the intensity of the workday can increase significantly, even if the overall workload does not decrease.”

Jones warned that increased productivity could lead to more stringent performance targets. Additionally, when routine tasks are automated, workers may lose the opportunity to gain the experience needed for more demanding tasks.

“So you could be expecting people to do more demanding work while at the same time reducing the experience that traditionally helps prepare them,” she says.

ArvatoConnect’s Maxwell argued that average handle times may need to increase as employees take on more complex cases.

“We assess whether[human]agents understand the customer’s needs, make sound judgments, and arrive at the right solution,” she said. “It’s not about whether they followed a script or hit some arbitrary goal time.”

Harris recognizes the benefits of AI tools, despite the occasional glitch in the system and increased workload. But she worries that increased automation will ultimately come at the expense of customer experience.

“I think as a country, we are focused on getting things done faster and choosing quantity over quality…As a result, customers are fed up with the way AI is being forced on everyone,” she said.

Shannon also recognizes the potential benefits of using AI, but believes it should not replace actual customer service.

“I think people buy things from people, but with AI you can’t have that experience,” she says. “I’d take it a step further. Can AI give you that feeling? Can it make it feel real?”