Abstract
While chatbots represent a potentially useful supplement to government information and service provision, transparency requirements imply the need to make sure that this technology is not confused with human support. However, there is a knowledge gap concerning whether and how government chatbots indeed represent a risk of such confusion, in spite of their resemblance with human conversation. To address this gap, we have conducted a study of a Norwegian municipality chatbot including interviews with 16 chatbot users and 18 municipality representatives, as well as analysis of > 2600 citizen dialogues. Interviews with citizen and municipality representatives suggested that citizens typically understood well the chatbot capabilities and limitations, though municipality representatives reported on some examples of humanizing the chatbot in its early phases of deployment. Dialogue analyses indicated that citizens have a markedly utilitarian style in their communication with the chatbot, suggesting limited anthropomorphizing of the chatbot.
