Human or Bot? How (Corrected) Errors Humanize a Communicator

Today more than ever before, online text-based interactions have become a common means of communication between consumers and companies. The current research investigates how one common feature in text communication—typographical errors (“typos”)—can humanize a communicator (e.g., a customer service agent). Across six experiments (N = 2,727) that used ambiguously humanlike conversational counterparts (i.e., cus¬tomer service agents who were either chatbots or real humans), participants perceived agents who made and subsequently corrected a typo to be more human than agents who made no typos or made but did not correct a typo. Moreover, perceiving an agent as more human led participants to form a more favorable perception of the agent (e.g., as being warmer and more helpful) and to be more willing to reward and engage with the agent. These findings provide novel insights into how conversational features influence customers’ perceptions of online agents, with implications for the agents’ corresponding brands. The current research also provides theoretical implications for anthropomorphism and human-computer interaction, as well as practical and ethical considerations for companies wishing to humanize their online customer service agents.