AI Takes the Routine Calls and Leaves Humans the Burnout

Contact centers were supposed to be the easy win for AI agents. The pitch was simple: let software absorb the repetitive calls, free human advisors for the hard ones, and everyone ends the day less tired. A CNBC investigation published on October 24 shows what is actually happening on the floor, and it is messier than the slide deck.

What changed inside the contact center

Some of it worked. Kaylee Harris, who handles customer service for a credit card company, told CNBC that her hours used to disappear into calls, note-taking and lengthy card disclosures. With AI handling the rote parts, she can spend more time on each customer.

Others describe the opposite. Shannon, a customer service worker at a telecom in Cardiff, said the AI-generated notes at her workplace are regularly wrong, so she rewrites them herself. At a previous employer, staff were told to "ChatGPT it" when they did not know an answer, while AI agents booked appointments and dumped every failed slot onto a human. "All the complicated cases came to us, causing a lot of burnout," she said.

The vendors are pushing hard in the other direction. Virgin Media O2 launched an AI voice agent in September aimed at routine issues. ArvatoConnect chief executive Debra Maxwell told CNBC the company targets 70% of interactions handled by technology and 30% by people, and that one of its tools drafts supplier viability assessments for the UK government's Crown Commercial Service with 97% accuracy. Nikola Mrkšić, chief executive of voice AI company PolyAI, frames the mission bluntly: contact-center staff often face "maximum abuse for minimum wage," and automation should carry part of that load.

The workload tax nobody priced in

The numbers suggest the load is not disappearing, it is being redistributed. MaxContact's 2025/26 UK Contact Centre KPI Benchmarking Insights Report found more than half of surveyed decision-makers said agent workloads had increased, up from around 40% in 2024.

Amanda Jones, who studies work intensification at King's College London, explained the mechanism to CNBC. Straightforward calls act as psychological respite between brutal ones. Remove them and the working day gets more intense even if total volume stays flat. Her second warning is worse: if juniors never handle easy cases, nobody builds the judgment needed for hard ones. Companies end up demanding more skilled work while destroying the training ground that produced it.

What builders should take from this

If you ship agents into any human queue, the handoff is the product. Escalations need full context attached, not a ticket number and a shrug. AI-written notes are a data pipeline, so measure their accuracy the way you measure deflection. And judge the humans on resolution and judgment rather than average handle time, or the automation will simply be graded into burnout.

The lesson generalizes well beyond call centers: automation does not delete the tail of hard cases, it concentrates it.