虎嗅APP·2025-10-19 13:20Core Insights - The article discusses the hidden labor force behind AI development, highlighting the disparity between the high valuations of AI technologies and the low wages of the workers who contribute to their training and evaluation [4][38]. Group 1: AI Workforce and Compensation - AI models require significant human input for training and evaluation, often relying on workers who perform tasks such as data labeling and content assessment [8][19]. - Workers in roles such as AI evaluators at companies like Google earn between $16 to $21 per hour, translating to approximately $3,000 per month, which is significantly lower than the salaries of AI engineers [22][23]. - The article emphasizes the irony of highly educated individuals, such as those with master's degrees or PhDs, being paid low wages for critical roles in AI development [21][25]. Group 2: Labor Conditions and Industry Practices - The work environment for data annotators is described as exploitative, with high expectations and low pay, often leading to burnout and job instability [28][33]. - The industry operates on a pyramid structure where a few algorithm experts are at the top, while a large number of underpaid workers form the base, creating a vast outsourcing network [30][36]. - The article points out that the reliance on low-wage labor for AI training is a global issue, with workers in various countries facing similar challenges, including psychological trauma from their tasks [37][39]. Group 3: Societal Implications - The article argues that the advancement of AI should not come at the expense of the dignity and respect for the labor force that supports it, drawing parallels to historical labor exploitation [38][39]. - It calls for a reevaluation of how society values different types of work, particularly in the context of technological advancements, to ensure that all contributors are recognized and compensated fairly [39].