Invisible Workers of the AI Ecosystem

Every technological revolution produces a class of workers whose labor is essential to the system's operation and invisible in the system's narrative. The AI ecosystem depends structurally on human work that its marketing materials never mention: data labelers in Nairobi, Manila, and Dhaka who categorize millions of images and text samples; RLHF trainers who rate AI responses for helpfulness; content moderators who review harmful outputs; cloud infrastructure technicians maintaining server farms; domestic workers (often spouses) who sustain household functioning while the builder builds. These workers make AI possible—without labelers, the models would not train; without moderators, the outputs would be unsafe; without domestic infrastructure, the absorbed builder could not sustain absorption. Yet their contributions are systematically erased, their names appearing nowhere in the celebratory accounts of AI's capabilities. The erasure is not accidental but structural: systems designed to maximize visibility of certain labor (engineering, founding, using) and minimize visibility of other labor (preparing, monitoring, sustaining).