Chinese Tech Giant Ties Employee Performance Reviews to AI Token Consumption

Chinese Tech Giant Ties Employee Performance Reviews to AI Token Consumption

A major Chinese tech company has begun linking employee performance reviews to AI token consumption, with token usage potentially affecting promotions, contract conversions, and career advancement. The policy, revealed by an employee on Zhihu, has sparked intense debate about AI adoption metrics and workplace surveillance in China's tech industry.

The Policy

According to reports on Zhihu (China's Quora-equivalent), the company's performance review system now includes AI token consumption as a metric:

• Employee token usage is tracked and quantified
• Token consumption levels could affect:

- 转正 (probation-to-permanent conversion)

- 晋升 (promotion decisions)

- 绩效评级 (performance ratings)

• Employees are encouraged or required to use AI tools in their work
• Token usage is measured across the company's internal AI platform

The Rationale

Company's likely reasoning:

• AI adoption is a strategic priority for Chinese tech companies
• Token consumption is a proxy for AI integration in daily workflows
• Higher token usage = more AI-assisted work = more productive/innovative employees
• Incentivizes employees to learn and use AI tools
• Quantifiable metric for management to track AI adoption across the organization

The Backlash

Employee concerns:

• "This incentivizes wasteful AI usage, not productive AI usage"
• Gaming the metric: Employees could generate meaningless AI queries to boost their numbers
• Quality vs quantity: Token consumption doesn't measure the quality or value of AI-assisted work
• Privacy concerns: AI queries may reveal sensitive information about work processes
• Mental health: Creates anxiety about using "enough" AI rather than using AI effectively

Industry reaction:

• Zhihu discussion reached 4.35 million views (significant engagement)
• Mixed reactions: Some see it as forward-thinking, others as dystopian
• Comparisons drawn to surveillance capitalism and Taylorism
• Concerns about AI token metrics becoming a form of digital piecework

Context: China's AI Push

• Chinese government has designated AI as a strategic priority ("New Infrastructure")
• Major tech companies (Baidu, Alibaba, Tencent, ByteDance) have invested heavily in AI
• China's tech industry faces intense competition in AI capabilities
• Internal AI platforms are common in large Chinese tech companies
• Employees are expected to embrace AI tools as part of digital transformation

Broader Implications

Metric design problems:

• Goodhart's Law: "When a measure becomes a target, it ceases to be a good measure"
• Token consumption is easily gamed and doesn't correlate with productivity
• Employees will optimize for the metric, not for actual value creation
• This could lead to widespread token waste without meaningful productivity gains

Workplace surveillance:

• Tracking individual AI usage adds another layer of employee monitoring
• Raises questions about trust, autonomy, and professional judgment
• May drive employees to use external AI tools to avoid company tracking

Global relevance:

• This type of policy could spread to other companies and countries
• AI adoption metrics are being discussed in management circles globally
• The balance between incentivizing AI use and measuring quality will be a key management challenge

What Would Be Better

• Outcome-based metrics: Measure the quality and impact of AI-assisted work, not token volume
• Quality audits: Review AI-assisted deliverables for actual value
• Training investment: Rather than penalizing low usage, invest in AI literacy training
• Voluntary adoption incentives: Reward teams that demonstrate innovative AI use cases
• Employee feedback: Involve employees in designing AI adoption metrics