Artificial intelligence may be taking over the workplace, but Meta employees apparently don’t have to be so enthusiastic about using it anymore. Meta has updated its performance appraisal guidelines so that employees are no longer evaluated based on how much they use AI tools. new WIRED report. The change moves the company away from the language of using artificial intelligence and puts the focus back on something much simpler: the actual impact of employees’ work.
It sounds obvious, but the distinction matters. Last year, Meta told workers that the “AI-driven effect” would affect their ratings, prompting employees to incorporate chatbots and AI agents into their daily work. Workers may also receive labels such as “AI Native,” “AI First,” or “AI Enabled” based on their adoption of the technology. The result, some employees say, has been a workplace where using AI can occasionally be a goal in itself.
The strange age of “tokenmaxxing”.
Some Meta employees pushed for AI tools more often, simply to increase their internal usage numbers. This included monitoring how many AI tokens, essentially pieces of information, were processed by the AI models. Things got so competitive that one employee created an internal leaderboard to rank colleagues based on their use of AI, complete with titles like “Token Legend.” The leaderboard was eventually removed after details about it were revealed earlier this year.
Meta now clarifies expectations. Engineers said this week that AI adoption dashboards and token numbers will not be used to measure their impact. The updated performance review language also says that employee outcomes are “supported by artificial intelligence or other means,” meaning that the use of AI is still highly recommended, but not necessarily the point. Meta spokeswoman Tracy Clayton told WIRED that the company has always evaluated workers based on their contributions, and said labels like “AI Native” were not used to evaluate performance.
Meta isn’t exactly slowing down on AI
Don’t mistake the change for Meta suddenly being less interested in hiring AI. If anything, the company’s workers will have access to increasingly effective AI tools. Employees recently tested an experimental AI agent called Hatch on their company devices. Unlike traditional chatbots, which mostly answer questions, Hatch can perform actions on the user’s computer, browse the web, and interact with other applications on behalf of the user. It has been available internally for several weeks and may eventually be released publicly.
This creates an interesting change in Meta’s approach. Employees are still encouraged to experiment with powerful AI systems, but simply accelerating the use of AI no longer counts as an outcome. Given that the Meta has reached the point where employees are competing to become “Token Legends”, judging work instead of token counters is probably not the worst reset.
\
