The human workforce that trains AI is like a vast, invisible scaffolding erected around the technology. It is absolutely essential for the construction of the final product, providing the support, guidance, and correction needed to shape the AI. But like all scaffolding, it is designed to be temporary and torn down once the building is complete.
This temporary nature is built into the employment model. Workers are hired as contractors, not employees, signaling that their role is contingent and short-term. Layoffs are frequent and expected, part of the process of dismantling one section of the scaffolding and moving on to the next phase of construction. There is no expectation of a long-term career.
This creates a workforce that is perpetually insecure. They are essential to the daily work of building the AI, yet they are treated as entirely expendable. This dynamic is a source of great stress and disillusionment, as workers realize they are building a technology that will ultimately have no place for them.
The great irony is that the industry’s dream of a fully autonomous AI—one that no longer needs the scaffolding—remains a distant fantasy. The technology continues to rely heavily on this human support structure. But the industry’s labor practices are based on the premise that this scaffolding is disposable, creating a system that is constantly undermining its own foundation.