Meta’s months-old AI unit is a soul-crushing gulag, say the engineers stuck inside it

Anyone who works at Meta or knows someone who works at Meta will tell you the same thing: it’s not a happy place, especially given the apparent endlessly layoffs the company has executed in recent years – cuts that have only accelerated as the company pours billions into AI.

Now, one new report in Wired suggests that the company’s Applied AI team is on the brink of revolt.

The drama started when someone hijacked an employee-only livestream presentation this week with an expletive-laden meltdown, demanding that attendees tell a senior Meta AI manager he was “a piece of shit.” One presenter reportedly covered their face with their hands.

The outburst, Wired reports, reflects simmering anger within the three-month-old unit of about 6,500 engineers and product managers tasked with supporting the company’s AI research ambitions.

Employees describe being forced into the group with no real choice: join or quit. Many call themselves ‘drawers’. Their assigned work? Generate puzzles and coding problems to train AI models. “It’s literally the gulag,” one employee told Wired. “Most people find the work crushing,” said another.

A report Last month in Business Insider, she shed light on how many employees learned they would be moved to the group — via a surprise email, a process that a self-described draftee later described on Reddit as “quite random.” According to an internal April announcement reviewed by Business Insider, Meta’s AI models still lacked the knowledge to outperform humans on technical tasks like coding. “To help agents understand how people perform everyday tasks using computers, we need to train our models based on real examples,” the post said.

In a leaked audio recording of an internal meeting that same month, Meta CEO Mark Zuckerberg explained the logic behind using Meta’s own engineers instead of outside contractors: Alexandr Wang – who sold his data labeling startup Scale AI to Meta for $14.3 billion before taking on the role of Chief AI Officer and leading Meta Superintelligence Labs – knows the world of data labeling well, and the company believes that the average Meta employee has “significantly higher” intelligence than that from third parties. contractors. It is then better to enable them.

Meanwhile, more than 1,600 Meta employees across the company have signed a petition protesting a program that monitors their clicks and keystrokes for AI training data. The mood across the company is so somber that Chris Cox, Meta’s chief product officer, felt compelled to address the “brutal” environment during a call with employees this week.

TechCrunch has reached out to Meta for comment.

According to previous reports, the Applied AI team is led by Maher Saba, a 12-year veteran of Meta who was previously a vice president in the Reality Labs division, the division that burned $83 billion on the Metaverse before Meta transitioned to AI. The new organization reports to Meta CTO Andrew Bosworth.

Originally it was set up in such a way that up to 50 employees reported to one manager.

Zuckerberg, for his part, reportedly addressed the situation in an internal memo on Friday, acknowledging that recent changes had “caused anxiety” and admitting that the company had made mistakes that it plans to address. According to Wired, he added in his memo that “Meta’s north star is the best place for the world’s most talented people to make an impact.”