Chen Dingding was a customer service representative at an e-commerce company — one of the first teams to be automated away when her company decided to deploy an AI chatbot system.

The announcement was framed as progress. The AI could handle thousands of queries simultaneously. It never took breaks. It never lost patience. It would reduce costs and improve response times. The company cut the entire team.

Chen received her layoff notice along with several colleagues. The severance was reasonable. The AI system was going to handle everything.

For about two months, it did handle a lot. Standard queries, order status, basic product questions — the AI moved through these efficiently.

Then the problems started.

It began with complex complaints. Customers who received damaged goods, who wanted refunds, who had been waiting for weeks for a resolution. When these customers interacted with the AI, it responded with templates. When that didn't work, it responded with more templates. When customers got frustrated, the AI offered helpful suggestions that were entirely irrelevant to their situation.

"遇到复杂的情绪类客诉，AI的回答要么是模板，要么是废话，客户越聊越气," Chen later explained.

Customer satisfaction scores plummeted. Escalations piled up. The AI had no way to read tone, to apologize sincerely, to offer a human gesture that de-escalated tension. It couldn't recognize when a customer wasn't angry about the product — they were angry about being treated poorly.

Then came the phone calls.

Customers who couldn't get resolution through chat started calling. The AI couldn't take calls. The phone queue grew. Supervisors were pulled in to handle cases the AI couldn't resolve. Meanwhile, the AI was creating new problems — its responses occasionally included incorrect refund amounts, its order processing had subtle errors that only showed up days later.

Two months after laying off Chen's team, her manager called her.

"回来上班吧。"

"Come back to work."

The AI boomerang is becoming a recognized phenomenon. According to Visier's analysis of 2.4 million employees across 142 companies globally, about 5.3% of laid-off workers have been rehired by their original employers — the highest rate since 2018. Gartner predicts that by 2027, half of companies that cut staff citing AI will rehire for similar roles.

Chen went back. So did about 40 other customer service representatives who had been let go from the same company's AI rollout.

"The work is different now," she says. "The AI handles the easy stuff — order lookups, shipping updates, basic questions. I handle everything that requires a human. It's actually better than before, in a weird way. I only get the interesting cases now."

She pauses. "But I also know: the next time the company decides it can handle this without us, I won't get another call. This is borrowed time."