Search

Developers of Artificial Intelligence (AI)-based systems are increasingly urged to assume accountability for their development decisions, referring to the degree to which they must justify underlying algorithms and their outcomes on demand. Thereby, AI developers often have to juxtapose how much accountability they self-attribute to them and how much accountability they perceive others attribute to them, creating intrapersonal perceptual accountability (in)congruence with unknown consequences for their job satisfaction. Building on perceptual congruence research and algorithmic accountability literature, we conducted an online survey of 87 AI developers about their experiences in AI-based systems development projects. Our results show that the lower the incongruence between self-attributed and others-attributed accountability, the higher the job satisfaction of AI developers. Moreover, we find that AI developers' role ambiguity mediates this effect. Our study contributes to a more nuanced understanding of AI developers' perceived accountability, with essential insights for defining job roles and understanding AI developers. [ABSTRACT FROM AUTHOR]

As accountability becomes increasingly important for developers of artificial intelligence (AI)-based systems, governance mechanisms such as AI principles or audits are often criticized for not sufficiently influencing AI developers. Therefore, we examine how visualized arguments in user interfaces (UIs) of integrated development environments (IDEs) can increase AI developers' perceived accountability. Combining construal level theory and Toulmin's model of argumentation, four UI design artifacts were developed, each containing a claim of process or outcome accountability with or without monitoring and evaluation tools that act as claim-supporting data. Results of an online experiment with 164 AI developers show that claiming process accountability increases AI developers' perceived accountability more than claiming outcome accountability, both without supporting data. However, when supporting data are available, both claims increase AI developers' perceived accountability comparably effectively. The study's results highlight the theoretical and practical usefulness of visualized arguments in UIs of IDEs to promote AI developers' accountability. [ABSTRACT FROM AUTHOR]

While designing artificial intelligence (AI)-based systems, AI developers usually have to justify their design decisions and, thus, are accountable for their actions and how they design AI-based systems. Crucial facets of AI (i.e., autonomy, inscrutability, and learning) notably cause potential accountability issues that AI developers must consider in their design decisions, which has received little attention in prior literature. Drawing on self-determination theory and accountability literature, we conducted a scenariobased survey (n=132). We show that AI developers who perceive themselves as accountable tend to design AI-based systems to be less autonomous and inscrutable but more capable of learning when deployed. Our mediation analyses suggest that perceived job autonomy can partially explain these direct effects. Therefore, AI design decisions depend on individual and organizational settings and must be considered from different perspectives. Thus, we contribute to a better understanding of the effects of AI developers’ perceived accountability when designing AI-based systems. [ABSTRACT FROM AUTHOR]

Accountability is crucial to make stakeholders of Artificial Intelligence (AI)-based systems justify their actions, thereby explaining the harm such systems cause to AI users. Due to the importance of accountability in the context of AI, accountability was introduced into IS research through literature reviews. Therefore, while IS research’s understanding of accountability covers the necessary depth, it comes at the expense of its essential breadth. Using a bibliometric analysis with 19,978 English-language papers, we shed light on the essential breadth posing three W- and one H-questions (When, What, Whereof, and How). Therefore, we contribute to IS research by highlighting the urgent need to revise existing definitions of accountability in the context of AI and establish them in IS research. We argue that a missing revision leads to non-transferrable findings within IS research. Accordingly, this study serves as a starting point for adapting definitions and creating a shared understanding in IS research. [ABSTRACT FROM AUTHOR]