Chatbots and Speech Act Responsibility Gaps

Abstract This chapter argues that chatbots create a new kind of responsibility gap, which it calls the ‘speech act responsibility gap’. Unlike the responsibility gaps commonly discussed in the context of self-driving cars and autonomous weapons, speech act responsibility gaps arise from the fact that paradigmatic speech acts like assertions (statements), promises, or orders always generate linguistic commitments and entitlements. Unlike more familiar kinds of responsibility gaps, speech act responsibility gaps are inherently interpersonal and directed. The chapter first argues that currently dominant treatments of chatbot speech acts as proxy agents cannot bridge these gaps. It also discusses why current arguments against the existence of responsibility gaps don’t apply in the case of chatbots. Instead, responsibility appears to be best attributed to the chatbot itself. However, this poses a dilemma. Either these machines don’t speak (we are fundamentally mistaken about their output) or we need to engineer (broaden) our notion of responsibility.