Half would have suspended a significant part of floor by which he meant reorganize your workforce around AI. As reported by Reuters, the project called Project OT, an acronym for Organization Transformation, plans to entrust an increasing share of daily activities to AI toolssimultaneously reducing the size of some work groups and the intermediate layers of the managerial structure.
Among the hypotheses evaluated by management, the workforce of certain teams could have been reduced by up to 60%. The downsizing would have been pursued through layoffs, failure to replace vacant positions and the removal of employees considered to be less performing.
Reuters reports that the overall reduction could have been comparable to that of 2023, when Meta cut about 25% of its workforce. Furthermore, in the original version of Project OT, a second wave of layoffs was also predicted in November. Zuckerberg would later decide to abandon, or at least suspend, that part of the plan.
One of the reasons for the revision appears to be related to results obtained from artificial intelligence within the company.
In the face of a 220% year-over-year increase in code changes relating to Meta’s AI platforms and infrastructures, the increase in development activity it would not have resulted in a proportional improvement in productivity. The new features, or updated ones, actually distributed to users would have only grown by 36%.
To this scenario a worsening of some operational indicators. Accidents of a technical and safety nature would have increased by 40%, while the time dedicated by employees to management of these problems would be grew by 70%. The data therefore suggests that the increase in activity generated or assisted by AI has not, at least so far, produced an equivalent increase in overall efficiency.
Zuckerberg himself acknowledged this publicly in a meeting with employees held in July. The CEO stated that the “trajectory of agent development over the last at least four months has not really accelerated as we expected,” adding that the company’s investments in AI agents “have not yet borne the desired fruit.”
Technological problems aside, Meta would also have had to deal with a deterioration in the relationship with employees. The company had in fact installed software on the computers of US employees capable of monitoring mouse movements and typing on the keyboard. The goal would be to provide future AI agents with enough data to replicate some of the tasks performed by employees.
The measure, however, provoked a negative reaction in the workforce, with indicators relating to employee “sentiment” reportedly decreasing by 19 points. At the same time, trade union organizing initiatives have also grown.
Meta has therefore not yet managed to transform huge investments in AI into measurable increases in productivity. However, suspending the plan does not equate to abandoning the overall strategy. According to Reuters, Meta continues to use smaller groups assisted by AI in some areas and is making significant investments to improve the infrastructure that powers its AI. Furthermore, Zuckerberg would not have ruled out targeted interventions on staff in the short term, leaving open the possibility of broader measures in 2027.
