The decision to lower quality barriers—and then effectively drop them altogether—would have sweeping downstream effects on the human labor behind AI systems. For years, the tech industry had relied on poorly paid workers in precarious economic conditions to perform essential data preparation tasks for its AI models, such as categorizing text and labeling images. Soon after GPT-3 normalized the use of giant, poorer quality datasets, the demands for the work shifted from the handling of largely benign content to frequently disturbing content, including for the purposes of content moderation,
...more
