How do data science practitioners experience and enact their emerging role in organizations while developing AI systems in-house, and with what consequences? This PhD dissertation answers the research question in three chapters. The first chapter examines the way data science practitioners navigate the paradox of explanations during the development of AI in organizations, revealing their active influence that these practitioners have on the level of transparency and opacity of their systems. The second chapter investigates how data science practitioners increase and decrease the complicatedness of their ML algorithms intentionally throughout the development process as a mechanism to redeem their identity as craft workers. The third chapter analyzes how data science practitioners enact their superior data science ethos while working in a support function in organizations, resulting in the creation of a proper space and status for their role. The dissertation contributes to understanding how data science practitioners directly and actively influence the development of AI in organizations in practice, as well as how they root for their role and push forward its importance within organizations. It highlights the need for managing efforts regarding how to handle and support data science practitioners in the in-house development of AI systems in organizations.
More information on the thesis