Select Language:
Leading major technology firms in China are now assessing the economic impact of employee token consumption, which has been steadily growing without delivering the anticipated benefits. In response, some companies have implemented restrictions and begun reviewing the results.
Initially distributing tokens equally among employees, these companies are now experimenting with more strategic and efficiency-driven approaches to allocate token quotas. Several employees reported that companies are quietly adjusting the limits on token distribution.
For example, a major tech company shifted from a uniform token allocation to a more flexible system managed by department leaders. This new approach considers factors such as individual needs, job roles, and other relevant criteria. As artificial intelligence becomes an integral part of daily operations, the process of managing token quotas has moved from broad distribution to more refined adjustments. Company leaders have emphasized that measuring token use is less about ranking employees and more about evaluating whether AI adoption enhances work efficiency and adds value.
Previously, one of the country’s leading tech firms viewed AI tool usage as a key performance indicator, equating higher usage with greater innovation. However, this perspective has shifted. An employee shared that once their token consumption hits half of their monthly limit, their supervisor reviews their usage and the actual results achieved during that period.
The motivation behind these changes stems from the realization that excessive token usage does not inherently improve operational efficiency. An employee from another prominent tech company explained that a team of around 20 members used about CNY50,000 (approximately USD7,380) worth of tokens in a month but didn’t see notable results.
Over the past two years, many companies have increased their budgets for AI-related initiatives. Yet, these investments were often driven more by fears of falling behind competitors or missing out on emerging technological trends than by solid business planning.
As organizations have gained experience, it’s become clear that token consumption doesn’t automatically translate into higher productivity. When companies treat token usage as a key performance indicator and invest heavily in AI without adjusting workflows and organizational structures, the true value of even the most advanced AI tools can become diluted.
Interacting with various firms has highlighted that some organizations only consider the explicit costs of API calls used for AI applications, overlooking the less visible expenses such as manual proofreading and data management. Others have used AI for low-impact tasks, which fail to significantly impact revenue or performance, resulting in increased token usage with minimal cost savings.
Evaluating AI effectiveness based on token consumption is a common practice among tech giants, and this metric is often linked to individual work assessments. While this encourages initial AI adoption, industry experts emphasize that the ultimate goal should be leveraging AI to enhance work efficiency and maximize organizational benefits.
ChatGPT
Perplexity
Gemini AI
Grok AI
