In recent years, many public and private organizations are starting discussions about the moral and ethical perspective on technological advancements of artificial intelligence (AI). The European Commission, for example, has published “Ethics Guidelines for Trustworthy AI”. In this collection they present requirements for the establishment of trustworthiness with AI artifacts based on human rights and fundamental ethical principles. Many other institutions have presented similar guidelines.
We investigate the design and effect of anthropomorphic capabilities of both, text- and speech-based chatbots. Especially, social cues and the emotional capabilities of Affective Chatbots are of interest. Such chatbot forms have potential use-cases for creating persuasion and nudging for unintended purposes. Therefore, we take this call to action and investigate how to design chatbots so that they support the users but do not harm anyone. At the same time they should create benefits for the community. We explore the fulfillment of these requirements through the implementation of trustworthy (Affective) Chatbots. This intention is based on the actionable application of ethical principles and based on real world problems we have experienced throughout our laboratory and field experiments.