Former Meta Employee Reveals: Highly Paid Engineers Are Actually Doing Data Labeling for AI
Former Meta employee reveals highly paid engineers are now doing AI data labeling as Big Tech restructures.
A former Meta employee has revealed how the company's radical AI-driven restructuring has transformed daily work: managers oversee 50+ reports via AI systems, lateral collaboration is nearly gone, and highly paid engineers are being assigned RLHF data labeling tasks. The revelations highlight a growing paradox in the tech industry — knowledge workers are effectively training the AI systems that may ultimately replace them.
When "Training AI" Becomes a Self-Eliminating Job
Recently, a former Meta employee revealed the real picture of AI-driven transformation inside Big Tech during an interview, sparking widespread discussion about whether "humans are training the AI that will replace them with their own hands." What makes this revelation thought-provoking isn't that it uncovers some earth-shattering secret, but that it paints an incredibly detailed picture of the organizational changes and career anxieties currently unfolding across the tech industry.
According to the former employee, Meta underwent a radical "flatten" restructuring internally: a single manager now directly oversees 50 reports, and lateral collaboration between employees has become virtually nonexistent. The daily workflow has been simplified to this — each person reports what they've completed to an AI system resembling an "open cloud," and the boss manages everything through that system.
The Technical Logic Behind Flat Management
Flat Organization refers to an organizational restructuring that reduces hierarchical layers and expands the span of management. Traditional tech companies typically adopt a 3-to-7-layer pyramid management structure, with each manager overseeing 5 to 10 direct reports. The aggressive flattening pursued by companies like Meta expands the management span to 50 people or more — something considered impossible in traditional management theory. The technological foundation for this shift is the integration of AI systems. Through automated workflow tracking, intelligent task allocation, and data-driven performance evaluation, AI effectively takes over the coordination, supervision, and information relay functions traditionally performed by middle managers. This model was first validated in Amazon's warehouse management and is now penetrating the knowledge work domain.
"You don't need to interact with people every day. You just tell the AI what you've done, and then the boss manages." This model may be impeccable in terms of efficiency, but as the whistleblower put it, it "makes you feel like there's really no joy in it." When collaboration between people is replaced by interaction between people and systems, the social dimension of work is completely stripped away, leaving nothing but cold task submissions and outcome evaluations.
Million-Dollar Engineers Doing Data Labeling for AI
Related articles
Design Intelligence and Engineering Insights from Singapore's MRT Information Display System
In-depth analysis of Singapore's MRT information display system's layered architecture, real-time data processing, and scenario-based design philosophy, exploring how public transit systems achieve efficient UX through engineering intelligence.
Natural Language Drives Blender: Generating 3D Scenes with AI Coding Assistants
Use AI coding assistants like ChatGPT Codex to drive Blender's Python API and generate stunning 3D scenes with just a few natural language prompts.
CAPI-DINO: A Combinatorial Approach to Adding Global Representations to Self-Supervised Learning
CAPI-DINO stacks DINO's global objective onto CAPI's local representations, achieving 68.7% linear probing accuracy on ViT-B/14 at only 16% compute cost.