Are Chatbots Human? Evaluating Potential Determinants of Anthropomorphism in Technology

Abstract

Anthropomorphism with regard to technology and artificial intelligence is becoming an increasingly relevant issue, due to the proliferation of modern technology across all age groups, while there still exists a great deal of uncertainty about its mechanisms and possible consequences. This study makes an attempt to contribute to the discussion by attempting to extend the three-factor theory of anthropomorphism (Epley et al., 2007) to the field of technology, in order to test whether this framework constitutes an effective model for explaining anthropomorphism in this context, by discerning whether an unpredictable chatbot is anthropomorphized to a greater degree than a more predictable one, as well as testing whether a user’s level of technological familiarity entails differing levels of anthropomorphism. An attempt to evaluate the effects of anthropomorphic language on technology is also made by employing emphasis framing (Druckman, 2001). These questions were tested in an experiment where the participants were asked to read a set of instructions, some of them employing anthropomorphic language, and then view a series of videos depicting a fake interaction between a user and a “chatbot”, after which the participants rated the interaction on several anthropomorphic measures. No statistically significant effects were found for predictability, framing, nor technological familiarity; however a significant effect with regard to the participants’ gender was found when examining the collected demographic data, which may constitute an as of yet unobserved effect of gender upon anthropomorphism.