Mensch-Algorithmus-Hybride als Quasi-Organisationen? Zur Attribution von Risiko und Verantwortung digitaler Kollektivakteure

Models of individualized responsibility for algorithms’ actions fail once one needs to focus on the human-algorithm-association itself as the unit of action. In some situations, human action and machine action are so closely intertwined that there is no linear connection between the emergent collectives and the complex interactions of the individuals from which they arise. In joint decision-making, the individual responsibility of humans or algorithms cannot be identified. Thus, a new perspective on human-algorithm associations is needed that recognizes their emergent properties, their distinct organizational character, and that develops appropriate models for a new type of collective accountability. The article addresses the following problems: How can the human-algorithm-encounter – within this socio-technical configuration – be theorized? Can the configuration be envisaged as a hybrid collective? Can it be perceived of in terms of collective agency? Which type of collective accountability will be appropriate for human-machine associations – centralized or distributed responsibility?