Wertbasiertes Design von Entscheidungsunterstützungssystemen - Von der Wertetabelle zur praktischen Umsetzung

The implementation of complex algorithms in decision support systems (DSS) offers companies new opportunities to reduce cognitive load in decisionmaking processes and to achieve better outcomes. However, voices from research and practice warn of the associated dangers. These include the systematic misjudgement of certain groups, incorrect assessment of (entrepreneurial) risks, and unclear responsibilities in the event of personal or material damage. Values are an increasingly important concept in this regard because algorithms and DSS based on them are not neutral artifacts, they are rather shaped by values and value assumptions arising during the development process. Value Sensitive Design (VSD) is used to account for this “value-ladenness” by creating a list of values that is considered important by users and developers with respect to a DSS. Except conceptual considerations, however, there is currently a lack of concrete practical operationalization of such value-lists. Using predictive policing as a practical example, we detail how a valuelist can be used for the design of a DSS and what difficulties can arise in this process. The employed three-step operationalization process provides clear guidelines to practitioners for considering the value-ladenness and resulting problems already during the development of DSS.