KJCKorea
Journal Central

This study examines whether generative AI can be regarded as a speaker from a pragmatic perspective, focusing on the growing reality of human–AI communication through natural language. It first reviews the concept of speaker in pragmatics and identifies four constitutive criteria: intentionality, context awareness, hearer-directedness, and interactional capacity. It then analyzes the language-generation mechanism of generative AI and related discussions in Korean linguistics and adjacent fields in order to evaluate its speaker-like functions. The findings suggest that generative AI cannot be recognized as an ontological speaker in the same sense as a human being. However, it does exhibit speaker-like functions at the level of functional intentionality, context sensitivity, hearer orientation, and interactional performance. In addition, users tend to experience AI-generated responses as utterances addressed to themselves rather than as mere informational outputs. On this basis, the study proposes the concept of the “functional speech actor” as a limited and conditional form of speakerhood.

This study examines the legal implications of using AI agents in academic research and proposes directions for institutional reform. Characterized by autonomy and interactive capabilities, AI agents are increasingly involved in all stages of research, transforming traditional human-centered processes into human– AI collaborative structures. This shift raises several legal concerns. First, copyright issues arise with respect to the authorship and originality of AI-generated outputs, making it necessary to distinguish AI-assisted creation from fully automated production on the basis of human contribution. Second, the use of training data creates challenges related to copyright infringement and the permissible scope of text and data mining. Third, responsibility in AI-based research is distributed across researchers, developers, and institutions. In response, this study emphasizes the need to maintain a human-centered copyright framework, clarify standards of originality, strengthen data governance, and expand researchers’ duties of verification, transparency, and research ethics in order to ensure responsible AI use in academic research.

This study investigates how university students conceptually understand generative AI and how such understanding shapes their expectations for its use. Written responses from 38 students on ChatGPT, its operating mechanisms, and future use were analyzed through the framework of mental models. The findings show that most participants relied on simple and intuitive rather than well-developed understandings of generative AI. Tw o misconceptions were especially prominent: the search engine model, which treats generative AI as a tool for retrieving and organizing external information, and the database model, which assumes that it recombines internally stored data. These mental models were closely related to students’ expectations, with many viewing generative AI primarily as an instrument for information search and organization. By contrast, students with more refined understandings envisioned broader applications, including learning, writing, idea generation, and problem-solving. The study highlights the need for higher education to foster both conceptual and practical AI literacy.