Socio-technical design of hybrid intelligence systems – The case of predictive maintenance

Abstract

This paper focuses on hybrid intelligence systems since this approach appears as most advanced to deal with the socio-technical challenges of Machine Learning. The analysis starts with a taxonomy being derived from literature that characterizes hybrid intelligence. This taxonomy is contrasted with the application area of predictive maintenance where machine learning could be used to derive warnings from the sensor data of a production plant. Thus, AI would replace the effort of deriving hypotheses based on the experience of the plant operators. The case study on Predictive Maintenance reveals a series of challenges that require keeping the organization and the human and in the loop. To achieve this goal, we derive appropriate design recommendations such as integrated support of communication and coordination, possibilities for intervention.