Abstract: Today’s artificial intelligence (AI), largely based on data-intensive machine learning models, relies heavily on the contributions of invisibilized and precarized ‘humans-in-the-loop’ who perform a variety of functions, between data preparation, verification of results, and even impersonation when algorithms fail. Using original data, I show that the ‘loop’ is one in which both machines and humans learn, but humans are at a disadvantage as they lack recognition and cannot build a personal and professional development trajectory. I conclude on the implications of these findings for AI ethics charts, which currently lack provisions to improve these workers’ conditions and quality of life.

Bio: Paola Tubaro is Research Professor in sociology at the National Centre for Scientific Research (CNRS) in Paris. A specialist of social and organizational networks, she is currently researching the place of human labour in the global production networks of artificial intelligence, and the social conditions of digital platform work. Her interests also include data methodologies and research ethics.