Meta has announced the termination of its tokenmaxxing incentive program for employees, which previously linked performance evaluations to the use of AI tools. The announcement was made internally this week, according to three employees who spoke to WIRED. The company is introducing a new AI tool called Hatch while employees report an increase in token consumption.
Previously, Meta had stated that employee evaluations would be based on their 'AI-driven impact,' which included metrics for chatbot and agent usage. This approach was criticized in a lawsuit filed by about two dozen employees, who claimed that the evaluation system violated US antidiscrimination laws during layoffs in May. The lawsuit alleges that employees on health and family leaves were unfairly penalized due to their inability to accumulate usage metrics, a claim that Meta denies.
The new performance review guidelines eliminate references to AI usage and the 'AI Native' designation, focusing instead on overall employee impact. Some employees have welcomed these changes, stating they feel less pressured to use AI inappropriately.
Meta spokesperson Tracy Clayton stated that the updates reinforce the company's long-standing practice of evaluating contributions rather than AI usage. Reports indicate that engineers were informed that AI adoption metrics would not be used for performance evaluations.
In recent months, some employees reportedly engaged in excessive use of AI tools to boost their internal usage metrics, leading to the creation of an internal leaderboard for tracking AI usage. Following the leak of this leaderboard, Meta restricted AI usage among employees.
Meta is currently encouraging employees to test Hatch, an advanced AI tool capable of autonomously performing tasks on computers. Concerns have been raised among employees regarding privacy and the potential for errors when connecting Hatch to personal accounts. Previous projects aimed at tracking employee activity have contributed to distrust in the company's AI tools.
Despite the cessation of the tokenmaxxing program, employees note that Hatch still consumes more resources than older AI tools. There are concerns that if Hatch successfully enhances productivity, it could lead to further job cuts at Meta, especially following reports of potential layoffs that were recently called off due to software issues and delays in AI development. Meta CEO Mark Zuckerberg has indicated that significant layoffs are not anticipated for the remainder of the year.