Micro-work, artificial intelligence and the automotive industry

This paper delves into the human factors in the "back-office" of artificial intelligence and of its data-intensive algorithmic underpinnings. We show that the production of AI is a labor-intensive process, which particularly needs the little-qualified, inconspicuous and low-paid contribution of "micro-workers" who annotate, tag, label, correct and sort the data that help to train and test smart solutions. We illustrate these ideas in the high-profile case of the automotive industry, one of the largest clients of digital data-related micro-working services, notably for the development of autonomous and connected cars. This case demonstrates how micro-work has a place in long supply chains, where tech companies compete with more traditional industry players. Our analysis indicates that the need for micro-work is not a transitory, but a structural one, bound to accompany the further development of the sector; and that its provision involves workers in different geographical and linguistic areas, requiring the joint study of multiple platforms operating at both global and local levels.