these algorithms can be held accountable. Opacity as to how algorithmic decisions came to be seems to be the norm and, while principles to which algorithms should adhere have been formulated, these lack proven methods that translate them into practice. Drawing on theories of design-science this research aims to fill that gap by the design of an artefact in the form of a checklist of machine learning monitoring methods that can be used to incorporate algorithmic accountability goals into decision-support systems. A qualitative research approach was taken where, after identifying algorithmic accountability goals from literature, experts in the field of data science were interviewed as to which machine learning monitoring methods could aid in the realisation of these goals. Findings from this stage were later validated using a focus group. Results indicate that the checklist, if embedded in an organisation in a similar strain as security or architectural principles, can aid professionals in the incorporation of algorithmic accountability goals in their decision-support systems.
- Algorithmic accountability
- machine learning
- machine learning monitoring
- ethical AI
- Master Business Process management & IT (BPMIT)