Meta retreats from plan to replace workers with AI

Meta Platforms ↗ 0.84% ambitious drive to remake its workforce around artificial intelligence has stumbled, with internal resistance, unreliable AI agents, and a sharp rise in site reliability emergencies forcing the company to pull back on several fronts.

The company's internal initiative, known as Project OT, aimed to cut team sizes by as much as 60% in two waves by replacing human workers with AI agents. But staff revolted, the agents underperformed, and CEO Mark Zuckerberg called off planning for further cuts hours before a major layoff in May, according to Reuters technology correspondent Katie Paul, who first reported on the plan and shared new details during a Reddit AMA on Wednesday.reuters

A Plan Derailed

Zuckerberg has told employees he got "the timing wrong," while CTO Andrew Bosworth acknowledged in an internal comment thread that the company did an "atrocious job explaining the vision" to staff. Major site reliability emergencies rose 40% year-over-year, and because AI-caused incidents were "weirder and less predictable" than human-caused ones, staffers spent 70% more time resolving them.reuters

Zuckerberg framed the layoffs to employees as a trade-off between two cost centers — compute infrastructure and people — telling them that spending more on one leaves less for the other. Meta already cut roughly 8,000 jobs, about 10% of its workforce, in a May layoff tied to its AI infrastructure spending. The company raised its full-year capital expenditure guidance to between $130 billion and $145 billion in its second-quarter earnings report in July.benzinga+1

Token Maxing Gets Retired

In a related reversal, Meta informed employees this week that it "will not use AI adoption dashboards or token counts to evaluate impact," according to an internal memo from executives Maher Saba and Santosh Janardhan. Managers were instead directed to assess "output quality, velocity, problem complexity and scope taken on".infoworld

The shift marks the end of a culture where engineers competed on leaderboards to consume the most AI tokens — a practice dubbed "tokenmaxxing." Meta found the gamification was diverting employees from regular work, as staff used AI to carry out unnecessary tasks to boost their scores. Bosworth had earlier warned employees in a memo that "nobody should be using AI tools just for the sake of using them" and that "token usage alone is not a measure of impact of any kind".infoworld

What Comes Next

Whether Meta has abandoned its workforce reduction plans or merely postponed them remains an open question. Zuckerberg has told staff he still believes the company's AI direction is correct and projected that AI models will improve enough to support his vision within three to six months. But he has also conceded that "the trajectory of the agentic development over at least the last four months hasn't really accelerated in the way that we expected".benzinga+1

As Paul noted in her AMA: "The cuts weren't abandoned out of principle. They were abandoned because the replacement wasn't ready".reuters