Will AI Replace Call Centre Jobs? What the Evidence Actually Shows (2026)

Short answer: not wholesale, and not the way the headlines suggest. But the entry-level rung of the ladder is being pulled up, and that is a more serious problem than the total-jobs number implies.

This article sets out what the evidence shows as of September 2026, with sources, including the findings that cut against the alarmist reading.

The scale of what's exposed

UK contact centres employ roughly 1.3 million people. Customer service and call centre work appears near the top of every credible exposure ranking, alongside administrative and clerical roles, because the work is routine, digital, and text-or-speech shaped — exactly what current models do well.

The automation figures are real. Gartner has projected conversational AI will reduce contact centre agent labour costs by $80 billion in 2026, and that 10% of agent interactions will be automated by 2026, up from 1.6%. AI chatbots and voice tools already handle an estimated 40–60% of customer queries in UK banking, retail and telecoms.

Note the shape of that: 10% of interactions automated, not 10% of jobs eliminated. Those are different numbers, and conflating them is how most coverage goes wrong.

What's actually happened to employment

Here the picture gets more interesting, and more uncomfortable.

The displacement is real but concentrated. A King's College London study analysing millions of job postings and LinkedIn profiles from 2021 to 2025 found firms with highly AI-exposed workforces reduced total employment by 4.5% on average — with the effect concentrated almost entirely in junior positions, which fell 5.8%.

Employer intentions point the same way. CIPD research found one in six UK employers (17%) expect AI to shrink their workforce over the next 12 months, rising to one in four among large private-sector firms. Of those expecting cuts, 62% identified clerical, junior managerial, professional or administrative roles as most likely to go.

The projections vary enormously. The IPPR's central scenario projects 545,000 UK jobs lost with GDP gains of 3.1%; its worst case reaches 1.5 million. The Tony Blair Institute projects 1–3 million UK jobs ultimately displaced, with peak annual displacement of 60,000–275,000 — against a historical UK average of 450,000 jobs lost annually to ordinary economic churn.

That last comparison deserves weight. Even the pessimistic AI projections sit below the rate of job destruction the UK economy absorbs every year without anyone calling it a crisis.

And the counter-evidence is substantial. PwC's 2026 AI Jobs Barometer found the most AI-exposed organisations grew headcount rather than cutting it — 52% growth against a 2018 baseline, compared with 36% at companies less able to use AI. The wage premium for AI skills rose to 34.2% in 2025, up from 11%.

Both things are true at once: exposed firms grow overall, and cut junior roles. That is the pattern, and it is not a contradiction.

What this means for a call centre agent

Routine, scripted, single-issue calls are going. Hours, order status, appointment booking, password resets, simple account queries. If your day is 80% of these, the job as currently designed will change substantially within a few years.

Complex, emotional and judgement-based calls are not. Complaints, vulnerable customers, disputes, anything requiring discretion or genuine empathy. AI escalates these, which means the human queue gets harder on average, not easier — a real workload change that most workforce planning ignores.

New work is appearing. Conversation designers, AI quality analysts, escalation specialists, prompt and intent maintainers. These are better-paid than the roles they replace. They are also fewer, and they require skills the displaced agent does not automatically have.

The entry-level problem is the serious one. Traditionally, call centre work has been how people without degrees enter white-collar employment and learn transferable skills. Automating the junior rung removes the training ground for the senior roles that survive. Nobody has a good answer to this yet, and it is a more consequential issue than the headline displacement figure.

What operators should actually do

Don't plan headcount reduction as the primary business case. The operations getting good results deploy AI on calls they were already losing — after-hours, overflow, the peak-hour spike — where there was never a human answering anyway. That is capacity recovery, not replacement, and it doesn't require a redundancy process.

Retrain deliberately, and early. Agents who understand the AI, review its transcripts and tune its escalation rules become the QA and conversation-design layer. That transition works when it's planned and fails when it's announced.

Redesign the human queue. If AI takes the easy calls, your agents now handle a harder mix all day. Adjust handling-time targets, quality scoring and break patterns accordingly, or you will lose good people to burnout while congratulating yourself on the containment rate.

Be straight with your team. The Contact Centre Association has published workforce transition guidance for a reason. Teams that learn about the AI deployment from a customer complaint do not cooperate with it afterwards.

FAQs

Will AI replace call centre agents entirely?

No credible evidence supports total replacement. The evidence supports substantial automation of routine call types, concentrated reduction in junior roles, and a shift in what remaining agents do.

How quickly is this happening?

Faster than previous automation waves, slower than vendor marketing implies. Interaction automation is moving from single-digit to low-double-digit percentages across a few years, not months.

Which call centre roles are safest?

Complex case handling, vulnerable customer support, escalation management, quality and conversation design, and anything requiring discretion or regulatory judgement.

Is it different in the UK?

The direction is the same. The UK's roughly 1.3 million contact centre workers, combined with a labour market showing the lowest vacancy levels since early 2021, means displaced workers face a harder time finding equivalent roles than they would have five years ago.

Should I leave the industry?

Not necessarily, but move deliberately toward the work that isn't automating: complex handling, QA, escalation, or the AI operations side itself. The people doing best in this transition are the agents who volunteered to review transcripts first.

Does AI actually do the job well?

For routine call types, yes — 70–80% automation rates are documented in production. For complex, emotional or novel calls, not reliably, which is precisely why those calls keep going to humans.

The bottom line

The honest summary is uncomfortable in both directions.

AI is not going to empty the call centres. Total employment effects look moderate, exposed firms are growing headcount overall, and even pessimistic projections sit below normal economic churn.

But the junior rung is genuinely being pulled up, the human queue is getting harder, and the industry's role as an entry point into white-collar work is eroding faster than anyone has planned for.

If you run an operation: deploy on the calls you're already losing, retrain early, and redesign the human queue for a harder mix. If you work in one: move toward the judgement work, and get close to the AI rather than away from it.

Sources: Gartner; King's College London (2021–2025 analysis); CIPD; IPPR; Tony Blair Institute; PwC UK AI Jobs Barometer 2026; ONS. Figures current as of September 2026.