Biopolitics, immaterial labor, and subjective dilemmas of China’s data annotators

This study adopts a biopolitical perspective to examine data annotation as a form of immaterial labor in China’s AI industry. Methodologically, it employed a mixed-methods approach in two phases: a three-month digital ethnography in a data annotation QQ group and two in-depth interviews in 2019, followed by twelve in-depth interviews with industry professionals in 2024. The analysis reveals how data annotators, now often termed AI trainers, experience intensified biopolitical exploitation masked as career advancement. Technological progress has reinforced the Taylorist logic of the work, binding laborers to standardized digital assembly lines. Capital has updated control via models like outsourcing, crowdsourcing, and remote work, further colonizing laborers’ life rhythms. Workers face alienation, fragmenting their subjectivity and leading to self-identification as ‘cyber slaves.’ However, some have begun embedding ethical judgments into AI training, transforming from governed ‘bare life’ into active subjects seeking voice. The paper argues data annotation is a manifestation of biopolitical production, generating not only commodified data but also exploitative social relations and labor subjectivities under capital’s biopower.