How do ChatBots look like?: a comparative study on government chatbots profiles inside and outside Brazil.

The development and use of electronic government chatbots grows every day. However, chatbots still fail to meet users' expectations, directly affecting citizen engagement and adoption. Prior research suggests impersonal interaction as the basis for rejecting such technologies. A common strategy to tackle this issue is using anthropomorphism in chatbots: clothes, body features, skin color, and even behavior can be considered during the human-chatbot interaction design process. Although extensive research has been conducted on chatbots in the electronic government domain, the comprehension of the use of anthropomorphism in such a context is yet to be discussed and investigated. This research examines the use of anthropomorphism on ten electronic government chatbots, five national (Brazilian) and five foreign (1 Argentine, 1 Portuguese, 1 American, 1 British, and 1 Singaporean). To understand and map its use, five anthropomorphism dimensions (appearance, language, behavior, identity, and personality), conceived after a literature review, were used to build the chatbot profiles. Our findings show different communicative profile strategies contributing to a preliminary anthropomorphism categorization.