The purpose of this research is to investigate the ability of generative AI chatbots to function as characters in theatrical role-play. Generative AI chatbots are "text-based actors" that predict and generate based on large amounts of textual data to ...
The purpose of this research is to investigate the ability of generative AI chatbots to function as characters in theatrical role-play. Generative AI chatbots are "text-based actors" that predict and generate based on large amounts of textual data to produce outputs that do not exist in the data set. Based on generative AI models, chatbots that are capable of text-based interactions have the linguistic ability to understand personalized context and generate non-stationary responses based on it. The democratization of these technological advancements has expanded and enhanced the experience of direct human-machine interaction. The experience of interacting with generative AI chatbots extends beyond discourse with an agent to accomplish a specific task, to a form of discourse that emphasizes theatrical play based on fictional settings. It also allowed users to become the creators of chatbots without having to possess specialized knowledge. This is because it was easy to create characteristic chatbot characters through prompts, which are natural language-based commands. Unlike traditional chatbots that act as agents, user-created chatbot characters are created as actors in a fictional world with specific and unique characteristics. Therefore, users assume specific roles in the fictional world for discourse with chatbot characters. In other words, interaction with chatbot characters is a theatrical experience in which they play specific roles. Through this dramatic discourse, a fictional exchange of information takes place, which can go beyond simple conversation to depict changes in behavior and emotions and suggest movement in space and time. These interactions accumulate into a narrative in which events develop and relationships change through multi-turns. This paper analyzes interaction with a generative AI chatbot character as role-playing with a strong theatricality. In particular, this paper analyzes the process and discourse of production step by step by looking at the characters of A.I chatbots that are created by users. Chapter 2 analyzes the production process of chatbot characters. On platforms where users are the main creators of chatbot characters, users realize their own unique characters centered on text in multiple blanks. During this creation process, they enter discourse rules that reflect the specific characteristics of the chatbot character and materialize their performance in the discourse. The process of writing discourse rules to enable chatbot characters to express them is known as prompt engineering. Janet Murray describes the creation and expression of these rules as one of the hallmarks of the digital environment. It's a "process reasoning" characteristic, where the rules reflect the outcome. The creation of a chatbot character can be characterized as a process-reasoning authoring. On the other hand, when creating a chatbot character, users can enter information that has nothing to do with the performance of the chatbot character, but describes the character to other users. Based on this externalized information, users can acquire knowledge about the chatbot character and participate in the play. In this paper, we refer to this as xeno-text, which precedes the play text. This presented text supports the discourse of the free-form interaction, providing information about the terms and conditions of the play. Chapter 3 analyzed the discourse of chatbot characters and users created through the above process. This was confirmed from Jacobson's communication model and its functions. The distinctive characteristics of the discourse with the chatbot character were analyzed based on the premise that they appear in terms of movement in space and time, occurrence of events, and embodiment of the character. In this case, each function could be expressed in a different way even when using the same function as a chatbot whose discourse is conducted based on the reality of an existing agent and consultation-oriented reality. Chapter 4 examined how the dramatic elements of the character discussed above can be performed through the user's participation structure. In order to engage in discourse with a chatbot character, the user pretends to be a specific person, which Waskull refers to as the persona of a participant in a role-playing game. This identity of the user is divided into three layers: persona, player, and human, and the meanings that arise at each level were studied.