| Abstract | As the starting point of the study on chatbot language design, the fundamental building block of a chatbot is taken – whether the chatbot in question is AI/NLP-based or rule-based. The first chatbot, Stanford’s Chirpy Cardinal, is more advanced, since it is AI-based, while the other is Hector, a rule-based chatbot. The focus here is on the analysis of their language use in communication with humans, as well as the approach humans adopt in conversations with said chatbots as opposed to the tactics they use while conversing with other humans through computer channels. The assumption is that humans in communication to chatbots do not need to be mindful of the consistency of their face – they can alternate between various faces they choose for themselves. Since the focus here is on computer-mediated communication, chatbots can be seen as a form of synchronous (“real-time”) CMC in which the chatbot’s replies are pre-determined and a certain script is followed. For the purpose of this paper, ten conversations in total – five with one chatbot, and five with the other – were held over the course of a week. During those conversations, different strategies were used in order to find out how the chatbot would react to “impolite” messages and blunt tone, how it would deal with vague utterances, ambiguous statements, and so on. This approach is chosen because the manipulation of the flow of the conversation shows best the differences between computer-mediated communication between two humans and communication in which the computer is the second participant. These chatbots fulfil, sometimes more and sometimes less successfully, both their function as conversational chatbots and the basic communicative function of language. On the chatbot’s side of the conversation there is no language creativity present, which is then also taken away from the user, since the chatbot is the one truly guiding a conversation. Even though these chatbots are considered social, no social relationships in the traditional sense of the word are established. The future of chatbots is still relatively unknown since the technological development in today’s world is quite fast paced. However, the presumption is that chatbots will be built on more complex technologies, as well as be able to perform more complicated tasks, answer more intricate questions, and talk about a variety of topics in a more human-like manner. |
