Other-ai

Meta's AI Unit Sparks Employee Outcry

Meta's three-month-old AI team of 6,500 engineers faces growing dissent, with some calling the work 'soul-crushing' and likening the environment to a 'gulag'.

Visual abstraction of neural networks in AI technology, featuring data flow and algorithms.

Photo: Google DeepMind / Pexels

A heated outburst during an internal presentation has sparked concerns about the working conditions at Meta's newly formed AI unit. An employee reportedly hijacked a livestreamed meeting, expressing frustration with senior leadership, calling the executive 'a piece of sh*t'. The incident highlights mounting tensions within the team, which has been under pressure to support Meta's AI research initiatives. The unit, which includes around 6,500 engineers and product managers, has been described by some as a 'gulag', with many feeling trapped and forced into the role with little choice. Employees have referred to themselves as 'draftees', emphasizing the lack of autonomy in their assignments.

The unit's primary task involves generating puzzles and coding problems to train AI models, a process that has been criticized as unfulfilling and mentally exhausting. A leaked internal memo from April revealed that Meta's AI models still lag behind human capabilities in technical tasks like coding. The memo stated that 'for agents to understand how people actually complete everyday tasks using computers, we need to train our models on real examples.' This sentiment was echoed in a leaked audio recording from the same month, where Meta CEO Mark Zuckerberg explained the rationale behind drafting internal engineers over external contractors. He cited the expertise of Alexandr Wang, who joined Meta after selling his data-labeling startup Scale AI for $14.3 billion, and claimed that Meta's employees possess 'significantly higher' intelligence than third-party contractors.

The situation has led to broader discontent, with more than 1,600 Meta employees signing a petition against a program that monitors their clicks and keystrokes for AI training data. The company's chief product officer, Chris Cox, addressed the 'brutal' environment during a recent employee call. TechCrunch has sought comment from Meta on the matter. The AI unit, led by Maher Saba, reports to Meta's CTO, Andrew Bosworth, and was originally structured to have up to 50 employees under one manager. In an internal memo, Zuckerberg acknowledged that recent changes had 'caused distress' and admitted the company had made mistakes it plans to address. He also reaffirmed Meta's goal of being 'the best place for the most talented people in the world to make an impact'.

Source: techcrunch

Key points

A livestreamed internal presentation was hijacked by an employee who called a senior Meta AI executive 'a piece of sh*t'.
Meta's three-month-old AI unit includes around 6,500 engineers and product managers, many of whom describe the work as 'soul-crushing'.
An internal memo from April stated that Meta's AI models still lack the knowledge to outperform humans at technical tasks like coding.
A leaked audio recording from April showed Mark Zuckerberg explained why Meta chose to draft internal engineers over external contractors.
More than 1,600 Meta employees have signed a petition protesting a program that monitors their clicks and keystrokes for AI training data.
Meta's chief product officer, Chris Cox, addressed the 'brutal' work environment during a recent employee call.
In an internal memo, Mark Zuckerberg acknowledged that recent changes had 'caused distress' and admitted the company had made mistakes.

Source: TechCrunch Read the original →

WRITTEN BY

Priya Anand

Emerging AI & Applications

Priya covers emerging AI applications and the wider impact of AI across industries.