Being proficient in AI has become a hard requirement in the workplace.

The rule of "Token consumption counts toward KPI" has spread from giants like Alibaba and ByteDance all the way down to small startups with just dozens of employees, issuing urgent notices pushing everyone to embrace AI.

Productivity has improved marginally—no one can quantify it. But workers' wallets have already shrunk.

Not every company is like Alibaba, which treats Token quotas as a free office benefit. Most bosses only care about results, not costs. To avoid falling behind or getting phased out, employees are forced to pay out of pocket, one after another, topping up their accounts.

Wallets Can’t Keep Up

It’s not even halfway through April, and Long Shen’s AI tool account is already running dry again.

Long Shen is a front-end programmer at a major e-commerce company, part of the first cohort of “AI-native” hires recruited in 2024. From day one, he began using AI to assist with coding. Last year, he started paying for AI tools.

His first payment went to Cursor—the hottest AI coding tool in the developer community. A monthly subscription on the official site costs $20, but an annual plan brings that down to $16 per month.

That $16 doesn’t buy unlimited access—it buys a monthly reset quota pool. Cursor charges based on actual Token consumption. When handling complex requests with multiple long-context conversations, the equivalent of $16 in usage can be exhausted within days.

The money was spent on work, yet there’s no reimbursement path. His company loudly proclaims “AI-driven efficiency,” filling internal emails with grand rhetoric about “intelligent transformation,” but when it comes to execution, no one ever mentions how Token quotas are allocated or how much can be reimbursed. Employees must foot the bill themselves.

Long Shen skillfully opens Xianyu, types “Cursor” into the search bar, and a list of listings appears: “clean accounts,” “quickly activated ready-made accounts,” “dedicated accounts.” Like a covert rendezvous, he clicks a link; the seller replies instantly: “Brand new dedicated account, refund proportional if banned within 30 days.”

Behind these links are mostly gray-market shared accounts or suspiciously sourced top-up credits. Sometimes Long Shen wonders: could this account be funded by stolen overseas credit cards?

He’s considered buying directly from the official site. But once work piles up, Token consumption flows like a river. To maintain output, his “tool arsenal” extends far beyond Cursor. ChatGPT Plus, Midjourney, various API services—he averages over a thousand yuan monthly, with peak months reaching 2,000 yuan spent across AI tools.

Paying to work, he tries to save wherever possible. After a brief hesitation, he still risks account bans and clicks “Buy.”

It stings to spend, but he calculates in his mind: spending an extra 1,000 yuan per month—only about 3% of his salary—helps complete 80% to 90% of coding tasks. The cost-benefit ratio makes the decision easy.

After paying, Long Shen’s workflow transformed completely. He took on a graphics-related project—high barrier to entry for most front-end engineers, rarely touched in daily work. Starting nearly from zero, he didn’t explain much to his manager but dove straight into AI-assisted development for three months.

“My boss doesn’t read code—he only cares if the page runs and functions are correct.” The project launched smoothly, earning Long Shen recognition. Only after stability did he slowly fill in the foundational knowledge he’d missed.

The company does offer a free internal coding tool. He tried it briefly but found it clunky. The tool only integrates domestic models, lacking cutting-edge core capabilities. It felt restrictive at every turn. After a few weeks, he abandoned it entirely and resumed paying for external tools.

He even tried promoting Cursor within his department. But once colleagues used up their free quotas, no one wanted to continue paying.

One colleague nearing 40 only rushed to him last month after the company mandated all-in AI adoption: “How do you use this thing? Teach me.”

Not everyone embraces the cost like Long Shen.

Sometimes I really think life would’ve been better without AI,” Pei Pei mutters while tallying her recharge expenses, caught between her company’s AI ban and her boss’s demands.

She works in design at an automotive manufacturer, where confidentiality rules are extremely strict—every external AI website is blocked outright. Accessing them from a work computer triggers a “connection failed” error.

Last August, things changed after her boss experimented with ChatGPT. Once exposed to AI-generated images, he dismissed all Pinterest and Instagram material as “secondhand”—already widely circulated online, making collisions inevitable.

He found AI-generated visuals inherently forward-looking, perfectly matching the need for bold, eye-catching designs. In meetings, he began directly demanding AI outputs, speaking as if it were just a finger tap.

Caught in the middle, Pei Pei uses personal devices to log into AI image generators, saves outputs to her personal email, then transfers them to the company computer for editing. This roundabout process is tedious and time-consuming—but there’s no alternative.

She gradually subscribed to Midjourney, Ji Meng, Ke Ling, learning each tool’s quirks. Her most-used tools are Dou Bao and Midjourney: Dou Bao is free and easy to use, ideal for simple color changes and basic adjustments, but its aesthetic is flat; Midjourney delivers high-quality visuals perfect for premium renderings, but it’s notoriously hard to control—changing one detail often ruins the entire image.

One month, she spent hundreds of yuan across several accounts. She tried requesting reimbursement from her boss, only to hear: “There’s no budget for that.”

She pays out of pocket, yet the workload keeps growing. Once bitten by the success of AI-generated designs, her boss’s appetite grows. Previously, two days’ buffer for revisions; now, he expects AI-powered efficiency to double output—today’s request, tomorrow’s version. Ten drafts? Then twenty.

“But people aren’t AI, nor machines,” Pei Pei grumbles, though deep down she knows: her boss only sees results, not effort, and certainly doesn’t care how much money she spends behind the scenes.

Sometimes she thinks: Edison invented the light bulb, but people didn’t enjoy easier nights—they just gained more nighttime work.

Once, her boss asked for a specific material rendering. She input the prompt repeatedly into AI, generating over thirty images, none fully meeting specs.

Finally, she turned off AI, opened Photoshop, manually stitched and color-corrected parts of several images. After over two hours of meticulous work, she dared submit the final version.

Membership Assassins Disrupting the Workplace

Li Huahua has lately grown increasingly paranoid.

Initially, AI’s emergence didn’t pressure her. As a programmer at a state-owned enterprise, her company enforces strict confidentiality rules limiting external tool usage. She treated AI as someone else’s problem—unrelated to her.

Until recently, late one night, a friend confided in her. Working at a private firm, he secretly subscribed to an AI membership to boost productivity. After delivering results, he proudly reported to his boss—only to be met with increased KPI targets. Now, everyone must do the work of two people.

After hearing her friend vent, Li Huahua sat in silence for a long time. Finally, she blurted: “Aren’t you exactly the kind of ‘code traitor’ people talk about online? Just chasing personal glory, dragging the whole team down?”

Her friend snapped back: “Then why don’t you start using it too?”

After hanging up, she couldn’t sleep. The next day, she spent the entire day researching how to subscribe to Codex.

But after activating the membership, she felt even more anxious. Her friend’s experience mirrored a warning: using AI for efficiency might not be a win. Maybe one day, she’d become the example used to justify raising KPIs—possibly leading to staff cuts. And recently, due to strained relations with her boss, she’d received low performance ratings for two consecutive months.

“Before using it, I feared being left behind. After using it, I fear everyone else is using it too. Constantly feel threatened, yet can’t pinpoint where the threat actually comes from.”

From then on, she began quietly observing her colleagues. Anyone whose work pace suddenly accelerated triggered suspicion: “Has this person secretly bought an AI membership too?” She never asked anyone—nor would anyone admit it if she did.

Li Huahua fears layoffs. Yet Long Shen’s company is now aggressively hiring AI talent.

Long Shen briefly participated in recruitment, drowning in resumes. The company explicitly required candidates to have real AI project experience and deployment cases. But sitting in the interview panel were veteran engineers with 15–20 years of experience—whose entire understanding of AI might boil down to letting their kids chat with Dou Bao about Ultraman.

With AI-driven efficiency, Long Shen gained more time to reflect—but realized the company was having non-experts guide experts.

Yet for management, this isn’t an issue. They hold big meetings, deliver presentations, break down KPIs layer by layer, leaving engineers to explore, produce, and report—all while they themselves needn’t learn or pay for memberships.

“They treat us like Agents,” Long Shen sighs helplessly. “Just give commands, consume our labor, and do nothing themselves.”

AI truly saved him time—but that time transformed into invisible labor: performing the appearance of being busy.

Now, he finishes his full day’s work in just a morning. To avoid drawing attention and triggering new assignments, he sits at his desk pretending to be occupied. Company computers are monitored—he dares not even connect to freelance gigs. Often idle, yet unable to leave.

This hollow feeling overwhelms him. His mind wanders uncontrollably: Should he invest in stocks? Buy gold? Is this how he’ll keep working until he’s optimized at 35?

He knows the AI advantage window is rapidly closing. In 2024, he could still leverage AI to stand out and earn recognition. By 2026, when everyone uses AI, individual differentiation becomes impossible.

Like when everyone starts attending tutoring classes—efficiency rises, homework increases, but no one gets to go home earlier.

In another major tech firm, programmer Zhang Mu finds himself trapped in a “AI-induced promotion trap” orchestrated by his boss.

One day, the department head suddenly dropped the March Token consumption leaderboard into the work group, declaring: “Promotion, KPI, and advancement will all reference Token usage. Low users may be replaced.”

Zhang Mu inexplicably landed at the top. The boss publicly praised him, assigning him to share “efficient AI usage” post-holiday. Instantly, he panicked: over half his Token usage was actually spent on organizing personal data, taking notes—tasks unrelated to work.

This put him on the spot. Forced to prepare a presentation, he dared not reveal his true, highly effective methods—those insights had taken weeks to refine. “I’m constantly afraid of being replaced. If I share them, my competitive edge vanishes.”

This pressure is spreading across the entire industry. Previously, people could barely manage with free tools like Dou Bao and Kimi—chatting, revising materials, handling daily tasks.

But that escape route is shrinking fast. Kimi began charging in September last year, starting at 39 yuan/month; Dou Bao launched paid tiers in Apple App Store this May—Standard: 68 yuan, Enhanced: 200 yuan, Professional: 500 yuan.

The era of “free assistant” is ending visibly. Want to use it? Pay up.

Can’t Stop Now

Before entrepreneurship, Jin Tu never imagined he’d spend so much on AI.

Having worked in brand marketing for years, like most, he used Dou Bao and Kimi for chatting, rewriting copy, and researching—enough to handle routine tasks.

Then one day, he saw a friend conversing with AI inside a code editor, realizing AI could directly generate documents locally—edit a version, save it, no more endless searching, copying, and pasting in chat windows.

After trying it himself, he unlocked a whole new world.

Since then, he began using AI for more creative and systematic tasks. He wanted to compile all his past public account articles into a knowledge base fed to AI. But WeChat’s anti-scraping measures made direct extraction impossible. He described the need to Codex, who delivered a custom browser plugin in just 2 minutes and 25 seconds. Open any WeChat article, click the plugin, and export it as a local MD file instantly.

Later, he built a personal knowledge workflow. Articles he reads, quotes, or excerpts get tossed in—AI automatically organizes them into structured notes, complete with his own analysis and commentary.

The most shocking moment came when his personal website was entirely built from scratch by AI—he wrote zero code. The site has since undergone 577 updates, attracting thousands of visits. Each update requires only one sentence: “Good, proceed.” AI then auto-checks, modifies, commits, and generates a detailed run log.

Thanks to this website, Jin Tu placed well in an AI startup competition and secured government-backed entrepreneurial support.

Maintaining this entire toolchain costs him significantly each month, but he believes it’s worth every penny. He quotes an AI entrepreneur: “Paying $200/month for Claude’s top-tier membership is like hiring a senior developer earning a million RMB/year.”

“Only after paying can you access real AI.” In his view, those unwilling to pay only experience stripped-down, discounted versions of AI. Using “real AI” is like buying a luxury bag—you know it feels different from a regular one, but can’t quite articulate why.

Currently, he’s planning his next move. Soon, he’ll head to Hangzhou to launch his venture.

Pei Pei still recharges her AI memberships occasionally.

Her boss specifically praised her growing proficiency with AI, urging her to keep it up. But she feels uneasy. Half of what AI generates doesn’t truly belong to her. Her inspiration is hers—but credit for the final image easily goes to AI. For designers, the recognition of having a proposal selected is everything.

What does her boss truly value—the idea, or the AI’s output? She can’t tell.

Li Huahua’s “tense heart” finally settled.

Even her nearly 50-year-old department head has recently begun preaching AI efficiency in meetings. Though he hasn’t yet said “one person doing two people’s work,” Li Huahua knows that direction is imminent. Now, she quietly maintains her membership, secretly uses AI, waiting for the moment to arrive.

Long Shen still buys accounts on Xianyu. Thanks to AI assistance, he’s been promoted three times in a year and a half, earned an A-level performance rating last year, and received nine months’ bonus.

This is AI’s true power: it subtly sweetens the deal, gradually eroding your workflow, making you willingly hand over money—and grow dependent.

After writing tens of thousands of lines of code with AI, Long Shen realizes he can’t live without it.

“I can’t possibly review tens of thousands of lines of AI-written code before taking over. Once this cycle starts, it’s nearly impossible to exit.”

Now, it’s no longer about whether to spend money—it’s about entrenched technical dependency. Maintenance can only continue via AI; stopping carries a higher cost than continuing to pay.

Author: Tian Mi, Image | Partial payment records of Long Shen, Image | After using AI, Li Huahua always feels under threat, Image | Website designed by Jin Tu using AI