The problem with the AI restructuring is that the AI wasn’t good enough.
Sometimes an AI-driven restructuring is accompanied by an AI-assisted legal filing. Reuters reports that Meta spent much of this year testing a plan in which AI would take over many of the staff’s daily tasks. Dubbed Project OT, the plan appears to connect two stories from earlier this year: Meta’s 10% workforce reduction and the company’s program that tracked employees’ mouse movements and keystrokes.
Meta’s leaders would have been influenced by startups, notably some they observed in Asia, which had been structured around AI. At an annual leadership retreat in January at Mark Zuckerberg’s Hawaii estate, executives laid out an “AI native” vision for the company. AI agents would be supervised by smaller “pods” of human employees. Engineers, designers, product managers and other specialists would increasingly move into general-purpose “builder” roles. Middle management levels would decrease and “agent-assisted analysis” would help determine daily priorities.
Executives have considered reducing the headcount of some teams by up to 60%. Layoffs would make up part of the reduction, as would eliminating vacant positions and excluding workers the company considered underperforming. A human resources executive predicts the reduction will be at least as large as the company’s job cuts in 2023, which reduced its workforce by 25%.
In its original form, Project OT (short for “Organization Transformation”) would have included a second wave of layoffs in November. CEO Mark Zuckerberg ultimately abandoned (or at least paused) this part of the plan, for reasons Reuters I didn’t find out. Meta acknowledged that there was initially a second plan, while presenting the more aggressive cuts as merely scenarios under consideration.
When technology can’t keep up
One reason for this change of heart may be that Meta’s internal AI wasn’t going so hot. On the one hand, the company’s data showed that code changes to Meta’s software platforms and AI infrastructure increased by 220% year-over-year. But that didn’t translate into equally dramatic productivity gains: new or improved features actually accessible to users increased by only 36%. And advances in these areas have been offset elsewhere, as the number of technical and security incidents has increased by 40 percent. Employee time spent on these issues increased by 70 percent.
In July, Zuckerberg appeared at a company town hall and admitted that he had overestimated how quickly technology would advance. He told staff that “the trajectory of agent development over the last four months at least hasn’t really accelerated the way we hoped” and that the company’s investment in AI agents hasn’t “paid off yet.”
Meanwhile, there was the tracking software the company installed on American employees’ computers. Surprisingly, capturing workers’ mouse movements and keystrokes to train their eventual AI agent replacements didn’t go as well. Complaints flooded Meta’s internal communications network, employee sentiment scores fell 19 points, and unionizing efforts intensified.
Combine the weaker-than-expected AI results with worker backlash, and perhaps why Zuckerberg ended this second round of layoffs isn’t so mysterious after all.
Always, Reuters says Meta hasn’t necessarily abandoned the larger plan. Zuckerberg’s message about further layoffs only reassured employees that further “company-wide” reductions would not happen “this year,” sparking internal fears about more modest reductions this year or broader ones in 2027. Smaller AI-powered teams remain in use in parts of the company, and Meta is spending heavily on AI infrastructure. As investors wonder when these investments will pay off, Zuckerberg will be under pressure to show the tangible gains that come with them.
