This paper uses an Australian case study called Robodebt to explore harms that were inflicted on social welfare recipients through the use of an Artificial Intelligence (AI) system. A socio-technical lens is used to identify the welfare recipients as social actors and to study their interaction with the Robodebt system. A dignity lens is then applied to view the impact on those social actors. Data were collected from written submissions by impacted welfare recipients to a Royal Commission on Robodebt. Narrative and content analysis methods were then applied to analyse the data. The findings conclude that harms would not have occurred if the perspectives of the dignity lens had been used in the design of the system. In designing future AI projects, it is recommended to consider individuals as multi-dimensional social actors and ensure that the protection of human dignity is a central consideration.