“Ghost Work: How to Stop Silicon Valley from Building a New Global Underclass,” by anthropologist Mary L. Gray and computer scientist Siddharth Suri, was published by Houghton Mifflin Harcourt in 2019. Based on years of fieldwork across platforms including Amazon Mechanical Turk, the book put a name and a human face on the invisible labor that powers supposedly automated systems.
Gray and Suri’s central argument is that the AI and automation we experience as seamless are quietly held together by millions of on-demand workers who label training data, verify outputs, moderate content, and handle the cases algorithms cannot. This work is deliberately obscured, hence “ghost work,” because the value proposition of automation depends on the humans staying out of sight. The workers are typically paid per task, managed by software rather than people, and left without the protections, benefits, or career ladders of conventional employment. The authors describe a “paradox of automation’s last mile”: every time a system automates one set of tasks, it tends to create new human tasks at the edge that someone has to do.
The book reframed the labor economics of AI. Rather than a story purely about machines replacing people, it revealed a sprawling, low-paid human workforce that is essential infrastructure for the industry. Gray and Suri close with proposals to make this work visible and dignified, an agenda that became more urgent as generative-AI systems greatly expanded the demand for human data annotation and content moderation.