Meta is officially ditching its "tokenmaxxing" incentive program for employees, meaning performance reviews will no longer hinge on how much they use AI tools. This shift comes as employees begin testing a new agentic AI tool called Hatch, which is already seeing a surge in usage.
Previously, Meta employees were evaluated based on their "AI-driven impact," which translated into how much they utilized chatbots and AI agents. This usage directly influenced internal labels like "AI Native," "AI First," or "AI Enabled." However, this system faced criticism, including a lawsuit alleging discrimination against employees on leave who couldn't accumulate usage and were thus penalized.
The new performance review guidelines replace specific mentions of "usage of AI" and "AI Native" designations with more flexible language, stating that outcomes "can be supported by AI or other means." This essentially returns the focus to overall impact rather than the specific method of achievement, freeing employees from feeling pressured to use AI when it's not practical.
A Meta spokesperson confirmed that the company has always evaluated employees based on their contributions and that labels like "AI Native" were never used for performance evaluations. Internal communications have also clarified that AI adoption dashboards and token counts will not be used to assess impact.
Despite the shift away from mandatory usage metrics, some employees have been actively prompting AI tools, sometimes frivolously, to maximize their internal token consumption. This trend even led to an internal leaderboard tracking AI usage, which was later removed. Meta previously attempted to ration AI usage after this practice became widespread.
Meanwhile, Meta is encouraging employees to test its advanced agentic AI tool, Hatch, which can autonomously perform tasks. While employees are testing Hatch on corporate devices ahead of a potential public release, some express privacy concerns due to its ability to access personal accounts and the company's past controversial AI data collection project. Despite these reservations, some employees are finding Hatch useful for personal organization, though worries about the environmental and economic costs of extensive token usage persist.
Although Meta declined to comment on Hatch testing, the tool is reportedly consuming more resources than previous AI assistants. Employees also suspect management still wants to see proof of their ability to effectively use AI. Some fear that if Hatch significantly boosts productivity, it could lead to further job cuts, despite recent reports indicating that plans for major layoffs have been postponed due to AI-related software bugs and development delays.
