AI usage targets in companies spark pressure and debate over performance metrics

Monitoring dashboards become standard practice in large corporations

Consulting divisions of major companies have begun implementing internal dashboards to track their employees’ use of artificial intelligence. At KPMG, the US-based services unit has set a 75% usage target for most of its approximately 10,000 employees, who can also compare their performance with that of their colleagues, according to HRKatha.

The initiative was introduced following staff cuts in the department and has already faced internal criticism, with some team members questioning whether usage frequency reflects meaningful work. At Meta, performance evaluations previously considered AI usage, but a recent change has altered this approach.

What happens when usage becomes an indicator

In September, Meta informed employees that performance evaluations would no longer consider AI usage, according to WIRED. The new guidelines state that results can be supported by AI or other means. The incident was linked to the term “tokenmaxxing,” which describes the practice of using more AI than necessary to improve personal performance metrics.

A survey conducted by Visier, a company specializing in people analytics, interviewed 1,000 full-time workers in the United States. Of these, 48% admitted to exaggerating their use or knowledge of AI in front of colleagues or managers, a behavior the report calls “performative AI.” Another 45% reported feeling pressured to use the technology without confidence in their ability to apply it correctly.

The most common requests from respondents were for training and greater transparency from leaders.

How to turn a usage requirement into real capability

The data raises a practical question: how can a usage requirement be converted into effective skill? Visier recommends that companies prioritize training and transparency before expecting productivity gains.

In practice, learning involves understanding what the tools do, where they tend to make mistakes, and how to review the results. The Pre-MBA course in AI offered by Saint Paul is an option for those who want to start learning about the topic.

Results over access

According to Andrea Derler, a researcher at Visier, adoption cannot be measured solely by the number of people using a tool or by the frequency of use. Without additional criteria, the dashboard may only record compliance with the rule.

The debate remains ongoing. Companies view dashboards as a means to track adoption, while employees argue that counting accesses does not demonstrate what has been delivered. The key question that remains is which indicators will be used to evaluate work performed with the support of technology.