This House would prioritize content moderation by human reviewers over algorithmic moderation.

tech

Meta's human-reviewer staffing has dropped by half since 2022. Trust-and-safety teams across the major platforms have been gutted. The work moved to AI; the failure modes shifted accordingly.

Background

Meta laid off roughly 21,000 employees in 2022-2023, with trust-and-safety teams disproportionately hit. X cut over 50% of its T&S workforce after the Musk acquisition. The AI-moderation systems now handling that volume have measurable strengths (extremism flagging) and measurable weaknesses (context-dependent harassment, sarcasm, breaking-news adjudication). The human-moderation literature also documents real harm: Sama and Genpact contractors in Kenya and the Philippines have reported PTSD-level trauma from reviewing graphic content for $1.50-3/hour. Both moderation modes have costs the other does not.

Government opens with

Algorithmic moderation fails at context (sarcasm, satire, breaking news) at precisely the moments stakes are highest.

Opposition responds with

Human moderation is unscalable, expensive, and traumatizing for the moderators; AI is the only path that respects worker dignity AND throughput.