China's Big Tech Forces AI on Workers, Sparks Backlash

Big Tech Turns Employees Into AI Test Subjects

Chinese tech giants are running controlled experiments on their own employees, splitting teams into AI-using and non-AI groups to measure exactly how much productivity AI can squeeze out — and potentially justify layoffs. The practice has sparked widespread employee backlash, with workers gaming AI adoption metrics in increasingly creative ways.

An employee at a major Chinese tech company, using the pseudonym 'Runsheng,' told Chinese tech outlet Zhiwei that his company divided existing projects into 2 groups. 'One group uses AI, the other doesn't,' he explained. 'The non-AI group gets a normal workload. The AI group gets assigned 140% of the normal workload — and it keeps increasing.'

The 140% Workload Experiment

Runsheng's concern cuts to the heart of the issue. 'It feels like the company just wants to see exactly how much AI can boost efficiency,' he said. 'And then the layoffs will probably come.'

This isn't an isolated case. Across China's tech sector, mandatory AI adoption has become a defining workplace trend in 2025. Companies are requiring all employees to embrace AI tools, tracking usage through formal KPIs and performance reviews.

The mandates typically include:

• Usage quotas: Employees must demonstrate minimum AI tool usage, often measured by token consumption
• Reflection reports: Workers are required to submit written reports on how they've integrated AI into their workflows
• Productivity benchmarks: AI-using teams face escalating workload targets, sometimes starting at 140% of normal capacity
• Comparative testing: Companies run A/B experiments pitting AI-augmented teams against traditional teams
• Performance reviews: AI adoption metrics are factored into annual evaluations and promotion decisions

Employees Fight 'Magic With Magic'

The response from employees has been predictably subversive. Rather than genuinely integrating AI into their workflows, many workers have turned to what Chinese internet users call 'fighting magic with magic' — using AI to game AI adoption metrics.

Zhiwei's investigation found employees employing several creative workarounds. Some workers simply use AI to write their mandatory 'AI usage reflection reports', satisfying the requirement with minimal effort. Others deliberately send AI tools to analyze massive open-source repositories on GitHub — feeding it tens of thousands of lines of code — purely to burn through tokens and inflate their usage statistics.

This behavior transforms what should be genuine productivity improvement into what Wang Anzhi, a professor at CEIBS (China Europe International Business School), describes as a form of corporate 'formalism' — the appearance of compliance without substance.

A Warning Sign for Western Companies

The phenomenon in China offers a cautionary tale for Western corporations pursuing similar AI mandates. Companies like Amazon, Google, and Meta have all made public pushes to integrate AI across their operations. Shopify CEO Tobi Lütke recently told employees that AI usage would be a 'baseline expectation' and factored into performance reviews.

The Chinese experience suggests that top-down AI mandates without genuine workflow integration can backfire spectacularly. When employees perceive AI adoption as a precursor to layoffs rather than a genuine productivity tool, they have every incentive to game the system rather than embrace it.

The core tension is straightforward: companies want to quantify AI's impact on productivity, but the very act of measurement distorts employee behavior. Workers who fear their jobs depend on proving AI's value — or who suspect the data will be used against them — won't provide honest signals about what works and what doesn't.

The Formalism Trap

This dynamic creates what organizational behavior experts call a 'measurement paradox.' The harder companies push to track AI adoption, the less meaningful the data becomes. Token consumption goes up, but actual productivity gains remain unclear.

For global tech leaders watching China's AI workplace experiment unfold, the lesson is nuanced. AI adoption mandates that feel punitive — particularly those paired with increased workloads or transparent efficiency benchmarking — risk generating resistance rather than innovation.

The most effective AI integration strategies will likely come from bottom-up adoption, where employees discover genuine use cases organically, rather than top-down mandates that turn AI into just another box to check. Until companies solve this trust gap, the risk of AI formalism — lots of token consumption, little real transformation — will remain a persistent challenge across the global tech industry.