The development of artificial intelligence has expanded in scope, allowing communication between artificial intelligence agents and humans possible. As for now, in terms of communication between artificial intelligence agents and humans, it goes beyon...
The development of artificial intelligence has expanded in scope, allowing communication between artificial intelligence agents and humans possible. As for now, in terms of communication between artificial intelligence agents and humans, it goes beyond the level of humans asking for information to artificial intelligence agents, which, in return give out answers. In fact, it has reached the level where the communication just for the purpose of itself takes place. An agent, that is a non-human being, does not only exist as an equal subject to the communication but also covers the difficulties inherent to communications between humans. Such phenomenon must be explained not in the light of how much the technology of artificial intelligence has developed, but more in the light of how agents communicate with humans and through which specific discourse, they make the talk happen. This study begins with the premise that discourse between artificial intelligence and humans resembles or models after the appearance of humans to humans discourse and selects chatbots, that are specifically designed for counselling among the ones that are released in app stores – chatbots including Stanford University’s Warbots, X2AI’s Sarah, LUKA’s Replika – as the subject matter. First, define the concept of those relevant chatbots and prove that those chatbots are the eligible candidates to make discourse. By analyzing the specific discourse of the subject matter, a unique characteristic of counselling based chatbots making discourse will be understood. The purpose of it is to consider that there is a post humanistic relationship between artificial intelligence agents and users. To further support the argument, based on the functions of language introduced by Roman Jakobson, a Russian structural linguist, the structural elements and functional models on discourse of chatbots’ service will be studied in the second chapter. Unlike other voice recognition services where relatively free flowing conversation is possible, chatbots are known to choose from the available options as in the form of closed discourse. And so, the study of Jakobson will be used to prove that discourse between chatbot agents and humans is still possible in the form of discourse itself. In fact, the study suggests that six elements in Roman Jakobson’s functions of language can be applied to chatbot services, which also correspond to the functions of language. In the third chapter, among many different chatbot services, chatbots specifically designed for counselling were studied closely to draw some characteristics of their discourse. First, it was confirmed that a series of interrogative questions were asked by chatbots designed for counselling – not the characteristic seen in those general commerce chatbots – indicating how much two-way message delivered through interrogative questions were intensified. This is to prohibit agents delivering the message all at once, unilaterally but encourages users to tell their stories. Second, an open discourse was possible with the message of feelings or affections fully delivered. Agents in chatbots designed for counseling convey more interactive expressions accompanying feelings or affections suitable under its ‘counseling’ platform. Since there is no premised purpose for the discourse, there is no clear or concise expression required as it is with discourse on commerce chatbots. This study looks at the aspect of discourse between chatbot agents and users through the specific statement. In the fourth chapter, the kind of relationship between chatbot agents and users forms is studied further through the analysis of discourse. According to the theory of Bruno Latour, French philosopher in the field of science, chatbot agents and users form mutually cooperative network through counseling sessions they experience given that the sessions are done non-face-to-face. Also, it was found out that users tell their genuine stories and thus, creating their own narrative self. Counseling based chatbot services do not take the role of an authoritative figure or a mere counselor but instead, placed on the same horizontal level with humans, listen to humans and respond with human-like emotions to form more interactive relationship with them. As such, this study seeks to contemplate on discourse that becomes the basis of what makes artificial intelligence agents and humans’ relation possible. It is deemed meaningful looking at this study considering how an authentic coexistence among two agents continues further.