AI in context: The labor of integrating new technologies

The report provides a holistic account of current challenges facing the integration of data-intensive technologies such as crop management tools and “smart” tractors in agricultural technology (a.k.a. “agtech”) and self-checkout stands in grocery stores. The authors detail the ‘human infrastructures’ necessary to integrate AI systems: Frontline retail workers must guide confused customers through the self-checkout stands or step in when new tech in the aisles fail. Farmers might have more agency in deciding to adopt new technology, but cannot fully realize the promised benefits of AI without leveraging resources to make crop data legible to machines. In order to acknowledge the tricky and sometimes counterintuitive ways that workers must labor to harmonize new “smart” technologies with preexisting processes, Mateescu and Elish propose these introductions be understood as integrations of automation and AI systems, rather than seamless “deployment.” Mateescu and Elish assert a perspective that rejects common AI tropes about robots and replacement in favor of a deeper examination of the everyday effects of AI on workplace responsibilities and conditions. They find that the integration of AI and automated technologies shifts physical infrastructure, processes, and norms in both workforces. Most importantly, the labor required to accommodate and implement these changes is undervalued and often rendered invisible. The report offers frames for those looking to develop, assess, and regulate automated and AI technologies.