AI Training Costs: Behind the Algorithm
Context
Your AI's brilliance comes from countless hours of data labeling, done by 'ghost workers' around the globe. Their wages are low, their conditions often poor. A local community group has started a viral campaign to raise awareness.
Dilemma
Do you:
A) Maintain current outsourcing practices, keeping costs low and profits high, or
B) Increase wages and improve conditions for data annotators, potentially impacting your bottom line, but gaining public trust?
Summary
Appen, an AI data service company, employs "ghost workers" globally to label data for AI training. These workers, often contractors, report issues like missed or delayed payments and poor communication. They perform tasks such as tagging images, transcribing audio, and categorizing text. Appen faces scrutiny over pay, with workers earning around minimum wage. The company claims it pays above location-based minimum wage. Workers also handle disturbing content, leading to psychological concerns. The work is described as essential but undervalued, with fears of job displacement as AI learns from their labor.
Resources:
| Last modified: | 10 April 2025 4.09 p.m. |