Join my new list, CaseBliz

Got put on hold for 22 minutes by my broadband provider this week. Transferred to a bot. Bot transferred me back to hold. I don't think the problem was a lack of artificial intelligence.

The narrative around AI and customer service jobs has always been one-directional. Automation arrives, headcount shrinks, boards celebrate. Except Gartner now predicts that 30% of employees laid off in AI-driven cuts across industries will need to be rehired by 2029, often at a higher cost. In customer service specifically, the correction is expected to hit harder and sooner.

Gartner's February forecast put the figure at 50% of companies that reduced service headcount due to AI will rehire by 2027. That's not a rounding error. That's a strategy reversing on itself within 24 months. Emily Potosky, Senior Director Analyst at Gartner, told CX Dive directly: companies that bet on flat-out AI replacement are going to have to rehire people to do the same work they tried to eliminate, while also carrying the cost of the AI they didn't cut.

The part worth sitting with: Forrester found 55% of employers now regret AI-driven layoffs. The jobs aren't coming back with the same title, though. Gartner expects "Customer Service Representative" to return dressed as "Solution Consultant" or "Trusted Advisor." The AI investment must be justified, even when it demonstrably didn't deliver. So the role changes, the salary probably doesn't, and the press release calls it an evolution.

Klarna is the example everyone reaches for, and fairly. The company cut 700 support jobs, watched customer satisfaction fall, and started rehiring. But Klarna's the one that talked about it publicly. Most companies are doing the same thing quietly, without a CEO blog post attached.

Gartner surveyed 321 customer service leaders and found 91% are under executive pressure to implement AI. The same research body found only 20% had actually reduced agent staffing because of it. That gap is the whole story. Boards are mandating AI adoption at a rate five times higher than the operational reality it's producing. The pressure is real. The promised displacement isn't, yet.

Every AI vendor pitch right now centres on the same promise: the more humanlike the AI, the better the customer experience. Build in a warmer tone. Add a name. Make it sound like it cares. Research published in the European Journal of Marketing in September found that, after service failures, customers interacting with a text-based chatbot reported greater feelings of relief than those using a voice bot. The less humanlike format actually performed better.

The explanation isn't counterintuitive once you hear it. Typing is slower and more deliberate than talking. When a customer types out a complaint, they have time to let the initial anger subside before the AI responds. A voice bot that sounds like a person but can't actually fix the problem is the worst of both worlds: it raises the emotional stakes without the tools to meet them.

The industry is spending enormous energy making AI sound more like humans. The smarter question might be: which format fits the emotion of the situation? Sometimes the right answer is a chatbot that doesn't pretend to be a person.

Re-reading the Gartner rehiring piece and thinking about the phrase "different job titles." The work comes back. The credit doesn't. Seen it before with offshoring.