Meta Project OT AI layoffs
por Huda Thomas: Mark Zuckerberg
This is Mark Zuckerberg's secret plan to replace thousands of Meta employees with AI workers. The internal documents just got revealed. And it explains every layoff this year. Here's what happened: In January, Zuckerberg and his top lieutenants gathered at his Hawaii compound for their annual leadership retreat. They came out of it with a plan code-named Project OT, short for Organization Transformation. The planning document describes an "AI native" future for the owner of Facebook and Instagram, where AI takes over much of the daily work performed by thousands of human employees. The humans who survive get a new job in that document: They "supervise“ the software. Virtual workers would be overseen inside Meta by smaller, "talent-dense" cadres of human staffers. The target was to cut many existing teams by 60%. Two layoff waves were scheduled, the first for May and the second for November. The May wave ran on time. Roughly 10% of Meta's engineers were kicked. Then between 20 and 30% of the engineers still on infrastructure and product teams were moved onto data labeling and AI training. So Zuckerberg pulled a quarter of his surviving engineers off building products and put them to work teaching the software that was scheduled to take the next round of jobs. And this is where it gets really interesting... On May 19, Zuckerberg killed the November wave. The entire plan rested on one assumption: That smaller teams running on AI would perform as well as the bigger teams they replaced. He ran that experiment with one of the largest technology budgets on Earth. And he stopped it HALFWAY. Then there's what Meta did four days ago: Engineers were told the company will not use AI adoption dashboards or token counts to evaluate impact anymore. For roughly ten months staff had been graded on their "AI-driven impact" and labelled AI Native, AI First or AI Enabled depending on how heavily they leaned on chatbots and agents. An employee had built an internal leaderboard called Claudeonomics that ranked the top 250 token burners and handed out titles like "Token Legend" and "Cache Wizard." It came down in April after it leaked. So people burned tokens they didn't need, because the number was the score. Meta measured AI adoption, and it got AI adoption. But it did not get what the Hawaii document promised. And in the same week the company dropped the token scoring, it started pushing an agentic tool called Hatch onto employee machines. Staff say their token consumption is surging anyway. About two dozen laid-off employees sued Meta in July, arguing the company broke US antidiscrimination law in the May cuts. Those AI labels show up in the case. Meta spokesman Tracy Clayton says labels like AI Native were never used for performance evaluation. Meta still posted record profits through all of this. It kept the 10% it cut. Wall Street applauded every announcement. And killing wave two might be about timing rather than surrender. But here's what should worry you: Every CEO in America is currently telling a board that AI lets them run a smaller team. Almost none of them have tested that claim. Zuckerberg tested it. He wrote it down, gave it a code name, put it on a calendar, ran half of it, moved his engineers onto labeling duty, and then pulled the plug in May. Nobody announced that last part. So when a CEO says AI means fewer people, are they showing actual evidence that it works?
Transcrição (en)
Meta explored cutting as much as 60% of the headcount from many of those existing teams. At the time, executives were really optimistic. Executives are imagining that with AI technology, a really talented person can do the work that was once performed by a bigger team. In the early days of planning, the scenarios that were being envisioned were not shared with employees. The restructuring was to be carried out in two waves, starting with a first one in May and then followed by a second shakeup in November. What Meta told us is that they got as far as determining what they wanted to do for the first wave, which took place on May 20th. So on May 20th, they laid off about 10% of the staff and they transferred roughly the same amount to new initiatives that they had spun up. Mark Zuckerberg called off planning for the second wave of restructuring the night before the May 20th layoffs. We don't know why he did this exactly. Employees by then had learned of what the company was planning. They saw signs that the company's AI transformation was at least partly aimed at replacing them, and they revolted effectively. They were posting angry comments on the company's internal workplace systems. They were sharing memes of elephants calling layoffs the elephant in the room. And so the company had a big morale problem on its hands, and executives were well aware of that.
