Meta CTO Admits AI Reorg Was 'Atrocious'

Meta's chief technology officer, Andrew Bosworth, admitted in an internal memo seen by WIRED that the rollout of the company's new artificial intelligence division was 'atrocious'. The comments come after WIRED reported widespread discontent within the Applied AI engineering unit.

Meta formed the division in March, comprising about 6,500 engineers and product managers, to enhance its generative AI models. However, employees complained about the menial nature of the work, with one describing it as a 'gulag'.

Bosworth wrote that the company had undermined trust that employees' expertise would be valued, that they would grow in their careers, and that they could make an impact. He cited disruptions to management structure and rapid strategy changes, including boom-bust hiring cycles, that left teams adrift.

To address the issues, Bosworth promised more personalized attention. Meta plans to cap managers at about 20 direct reports and limit how often employees switch managers during restructurings. Managers will focus primarily on people management rather than individual work. Employees will also have access to 'AI coaching' tools.

In a separate post, vice president Maher Saba told employees who were forced into the AI team that they could now apply for other roles within Meta.

Bosworth emphasized that Meta does not believe AI will fully replace workers, but noted that 'someone who knows AI might take your job'. He acknowledged tough trade-offs around compute resources and pledged transparency.

To boost morale, Bosworth promised to make Meta a 'fun and enjoyable' place by improving office microkitchens, increasing travel budgets, and spending more on social events for in-person bonding.