Two years ago, Wei was the kind of employee who bragged about his hours. Nine to nine, six days a week — the infamous "996" schedule of China's tech industry. Emails, code, meetings, presentations, data reports. He was busy, he was needed, and that was his security.

Then AI arrived — not with a bang, but piece by piece.

It started with weekly reports. He'd throw in a few keywords, and the AI would generate a polished summary. Saved him half an hour. Nice.

Then his boss started sharing AI tool links in the group chat. "Everyone try these, improve efficiency." He did. Emails, meeting minutes, English translation — the AI handled it all. He started depending on it, but told himself it was just a tool upgrade, like switching from a typewriter to a computer.

The acceleration began at the start of 2026, when the company brought in several AI platforms connected directly to internal systems. Data reports that used to require manual pulls? Auto-generated and pushed on schedule. Proposal documents he used to refine word by word? First drafts from a few keywords. Requirement descriptions that needed back-and-forth with designers? High-fidelity prototypes in seconds.

By March, he was working about three hours a day. The rest was scrolling, staring, pretending to be busy. By May, it was down to two hours. By July — one hour of real work, max.

"I sit at my desk, screen on, fingers on keyboard, looking busy," he wrote. "But inside I'm screaming: AI is taking my job one piece at a time, and I can't even fight back — because it's doing it better."

The worst part is the metric. His company has started tracking everyone's "AI efficiency improvement rate" — how much time AI saves you. His number: 83%. On paper, it's a win. In his head, it's a countdown. What used to take him a week now takes one day. The question that keeps him up at night: what about the other four days? What's his reason for existing?

He did the math himself. The company spends about 20,000 RMB per month on him — salary, social insurance, housing fund, office costs. The AI tools replacing him? Less than 1,000 RMB a month. "If I were the boss, why would I keep me?"

He doesn't dare talk about it with colleagues. Bring up the topic and the room goes quiet. Everyone pretends everything is normal, everyone pretends their workload is just as full, everyone pretends AI is just an assistant. But everyone knows the real game: who gets cut first.

The scariest realization came one evening after work. He sat in his car, opened the AI tools on his phone, and tested them against everything he used to do. Write code — yes. Write documents — yes. Data analysis — yes. Even his prized "cross-departmental communication skills" — the AI generated flawless email drafts.

One by one, they passed. He had to admit it: five years of building his "core competitiveness," and it crumbled in front of AI. "My experience is its training data. My skills are its standard features. My prized 'efficiency' — just one line of its code."

He's trying to find a way through — focusing on work that's hard to describe, on understanding the "why" instead of the "how," on being the connector between business and technology. But he knows these are just temporary safe zones.

"Security never came from what you know," he wrote. "It came from what you can still learn. But even that sounds tragic these days — because AI learns a million times faster than we do."

At the end of the essay, he revealed that the AI assistant had offered him title suggestions and template structures for this very article. He closed it and wrote it himself.

"This article, I want to write myself. Maybe it's not as good, maybe it won't get many views, but I want it to be mine. Not because AI can't do it, but because I want to."