AI technology is increasingly being adopted in practice for all kinds of tasks. To apply AI in organizations, it must be properly designed and executed. To guide and assure proper design and execution, principles are formulated. Currently, the body of knowledge on principles for the design and execution of AI technology is a growing field, where contributions build on principles that represent ethical values from other, more mature, research domains such as the medical or bio-ethics field. However, many of those principles are general and unusable for the design and execution of AI. In this study, we aim to formulate design principles to guide the design and execution of AI technology in practice. This is achieved by an inductive approach in which guidelines on AI technology design and execution of large international organizations are analyzed, involving five phases of coding with five researchers. The analysis resulted in the identification of 22 principle categories that represent different ethical values such as ‘accountability’, ‘understandability’ and ‘equality’. Based on this, design principles are defined for each principle category. The results of this study form a basis for making AI design principles more concrete, so that practice can actually apply the design principles to improve AI design and execution. Additionally, the results show the importance of value-sensitive design of AI.
Smit, Koen; Zoet, Martijn; and van Meerten, John, "A Review of AI Principles in Practice" (2020). PACIS 2020 Proceedings. 198.
When commenting on articles, please be friendly, welcoming, respectful and abide by the AIS eLibrary Discussion Thread Code of Conduct posted here.