For Kaylee Harris, who works in customer service at a credit card company, today's workdays seem quite different from before.

She told CNBC that in the past, her working hours were often filled with answering calls, taking notes, and dealing with lengthy credit card disclosure forms. Now, with the help of artificial intelligence, she can devote more of her energy to each individual customer.

For Shannon, AI also changed everything – but not in a positive way. She works in customer service at a telecommunications company in Cardiff, Wales, and due to concerns of being recognized by her employer, she prefers not to disclose her surname. She told CNBC that at her current workplace, the records generated by AI are often full of errors, “so I simply write them myself.”

She said that at her previous employer, a property service company, senior management would tell employees to “use ChatGPT to check” if they didn’t know the answer. The AI customer service agents were responsible for making appointments, but if they couldn’t provide the date requested by the customer, they would transfer the case to manual processing.

Shannon said that employees are 'always putting out fires.' 'All the complex cases end up in our hands, leading to severe burnout.'

Is it beneficial for both employees and customers?

Theoretically, automation in customer service should mean that human representatives spend less time on administrative tasks and devote more time to helping customers with issues that require human judgment, empathy, and experience.

Šić, the CEO and co-founder of PolyAI, which develops customer service voice AI systems based in London, stated that contact center agents may face "the lowest salary and the highest level of abuse."

He said to CNBC, these employees often have to deal with angry customers, and at the same time, they work under strict performance indicators. Sometimes, they don't have the authority or enough information to solve the problems. He believes that AI customer service agents can help share this burden.

"This is truly a very, very difficult task," he said, "I believe that with AI, we have the opportunity to bridge this gap and make it better for both parties – customers and employees."

British Telecom, in collaboration with media company Virgin Media O2, launched a AI voice agent in September. The company stated that this agent is designed precisely to achieve this goal: to handle routine inquiries, thereby giving human consultants more time to deal with complex or sensitive cases.

The CEO of a UK customer experience company, ArvatoConnect, Debra Maxwell, stated to CNBC that the company's goal is to have 70% of interactions handled by technology and 30% by humans.

She said that a tool AI developed by the company for the British government procurement agency Crown Commercial Service can generate draft supplier feasibility assessments with an accuracy rate of 97%, thereby reducing several hours of manual data extraction work.

The psychological burden brought by more difficult cases

As AI takes over routine tasks, human customer service staff may be left to handle a larger proportion of complex cases—Moreover, they will also have to clean up the mess when technology fails.

As reports of increasing workloads across the industry continue to emerge, the pressure on customer service staff is also on the rise.

According to the "2025/26 UK Contact Center KPI Benchmark Insight Report" published by MaxContact, over half of the contact center decision-makers surveyed indicated that the workload for agents had increased, which is higher than the approximately 40% in 2024. However, this survey does not explain the reasons behind this rise.

The research organization at King's College London that studies organizational behavior and future work, Amanda Jones, specializes in the study of “work intensification,” which refers to the process by which work becomes increasingly demanding over time. She stated to CNBC that if human operators are increasingly forced to deal only with the most difficult interactions, AI could potentially make call center work even more intensive.

"Some customers may already feel frustrated before they even get in touch with a human representative," she said. "What we sometimes overlook is that relatively simple and straightforward tasks can provide a psychological respite between more demanding assignments. If these tasks disappear, even if the overall workload doesn't increase, the intensity of each workday could still rise significantly."

Jones reminds us that increased productivity may also come with more stringent performance targets. As routine tasks are automated, employees may also lose the opportunity to gain experience, which is essential for being competent in higher-demanding jobs.

She said, "Therefore, we may end up asking people to perform tasks of higher standards, while at the same time reducing the experiences that traditionally help them prepare for such tasks."

The temperature of a person

Maxwell of ArvatoConnect believes that as employees take on more complex cases, the average processing time may need to increase.

She said, "What we evaluate is whether the human representatives have understood the customer's needs, whether they have made reasonable judgments, and whether they have arrived at the correct solutions, rather than whether they have just recited from a script or whether they have met some arbitrary time targets."

Despite the occasional system failures that actually create more work, Harris still sees the benefits of the AI tool. However, she is concerned that the trend towards automation is ultimately being achieved at the cost of the customer experience.

“I think, as a country, we’re focusing on doing things faster, choosing quantity over quality… As a result, customers are tired of the way AI is being forced upon everyone,” she said.

Shannon also sees the potential benefits of using AI, but she believes that AI should not replace real customer service.

“I believe people buy things from people; with AI, you can’t experience that,” she said. “I would go the extra mile. Can AI give that feeling? Can it seem real?”