In the world of AI training, everything is reduced to a number. The value of a worker, the quality of their output, and the progress of the technology are all measured in clicks, tasks per hour, and other cold, hard metrics. This relentless, metrics-driven approach is at the heart of the human cost of AI, creating a system that is efficient at generating data but terrible at nurturing human well-being.
“Just get the numbers done” is the mantra that governs the daily lives of AI raters. This singular focus on quantifiable output forces them to make a constant trade-off between being thorough and being fast. As one worker learned, trying to “get things right” and truly understand the task is a surefire way to be flagged for low productivity.
This system is particularly damaging when applied to knowledge work that requires nuance and judgment. A metric can tell you how many responses a worker rated, but it can’t tell you if the ratings were thoughtful, accurate, or ethical. By optimizing for the things that are easy to measure, the industry is systematically devaluing the things that actually matter for quality and safety.
The human cost of this obsession with clicks is immense. It leads to burnout, anxiety, and a profound sense of alienation, as workers feel their expertise and conscience are irrelevant. They are reduced to human bots, valued only for their ability to generate data points at a rapid pace, a dehumanizing reality at the core of our most “human-like” technology.
