“Why aren’t you getting this done? You’ve been working on this for an hour.” This question, posed by a supervisor to an AI trainer, perfectly encapsulates the impossible demands placed on the human editors of artificial intelligence. An hour is seen not as a reasonable amount of time for a complex task, but as an unacceptable delay in a system built for hyper-speed.
The trainer in question was trying to do her job diligently, taking the time to understand new guidelines and apply them correctly. Her reward was not praise for her commitment to quality, but a reprimand for her low productivity. This experience is typical in an industry where the clock is the ultimate measure of performance.
These impossible time constraints force workers to make compromises. They can’t afford to double-check every source. They don’t have time to craft the perfect, nuanced edit. They must constantly choose between doing the job well and doing the job fast, and the pressure from management always pushes them toward the latter.
This has a corrosive effect on both the workers and the product. The workers suffer from stress and a sense of professional failure, knowing they are not doing their best work. The AI, in turn, is trained on a diet of rushed, superficially reviewed data, making it less reliable and more prone to error. The demand for impossible speed is creating a foundation of inevitable failure.