Is Meta destroying its engineering organization?
Meta’s aggressive pivot to AI — including reports of mass reassignments of core engineers to data labeling and RLHF work, heavy internal tracking, and recurring layoffs — is seen by many as gutting a once-strong engineering culture. Commenters debate whether using highly paid software engineers as “annotation labor” is smart frontier AI investment or a wasteful, demoralizing way to force attrition in an over‑staffed ad business. The broader concern is that AI hype and crude metrics like “token leaderboards” are driving irrational management decisions across big tech, with long-term risks for product quality, worker morale, and the industry’s future.
Scale of AI Reassignment and Layoffs
- Thread centers on claims that 30–50% of engineers on some “core” / infra teams were moved into AI data-labeling / RLHF work, plus ~10% company-wide layoffs.
- Multiple self-identified insiders say those percentages are plausible or even higher for specific infra teams; others find them “unbelievable” but concede leadership might see it as rational.
- Many view the moves and constant layoffs as a way to force attrition and shrink headcount without explicit mass firings.
Why Use Expensive Engineers as Labelers?
- Pro argument: frontier RLHF requires deep domain expertise, especially for coding agents. Detailed multi-turn annotation and ranking are mentally taxing and hard to offshore cheaply.
- Counter: US FAANG engineers are overkill; similar or better quality could be sourced in cheaper markets or from specialized labeling firms. Forced, resentful labelers are unlikely to produce “high quality” data.
- Some frame it as “soft layoff” or “training your replacement” work that won’t last.
AI Psychosis, Metrics, and Surveillance
- Many see a broader “AI psychosis”: top-down mandates to “AI-ify everything,” token leaderboards, and massive internal LLM usage (“tokenmaxxing”) used as performance proxy.
- Heavy monitoring (screen/keyboard tracking for AI training and productivity) is described as invasive and dystopian; others reply that privacy on work devices was never real anyway.
- Fear that leadership is chasing AI fads in a panic, mirroring earlier failures like the metaverse.
Impact on Culture, Morale, and Org Health
- Reports of managers pushed back to IC roles, chaotic reorgs, and leadership paralysis. Some teams allegedly lost half or more of their engineers to the AI org.
- Internal attrition said to be high and rising; many plan to leave after next vest. Internal comms reportedly shifted from aggressive AI push to pleading for people not to quit.
- Several commenters argue Meta’s core social products could be maintained by far fewer engineers; over-hiring plus perf-obsessed culture made this kind of purge inevitable.
Ethics and Reputation
- Long, heated debate about the morality of working at Meta given alleged harms: addictive design, teenage mental health, scams, propaganda, and enabling atrocities.
- Some argue employees share real moral responsibility; others emphasize economic necessity, lack of “clean” employers, and the difficulty of drawing bright ethical lines.
- Broader worry that this AI-heavy, metrics-driven management style may spread across the industry, not remain confined to Meta.