This phenomenological and sociolinguistic study investigates current human perceptions of dialogue with generative artificial intelligence (AI), focusing on intellectual, emotional, and moral dimensions. Participants were female humanities and social sciences researchers in their thirties with extensive AI experience. Data were obtained through in-depth interviews based on real-time interaction with ChatGPT and analyzed using Moustakas’ (1994) framework. Results indicate that AI was construed as a strategic collaborator restructuring information into research insights and as a creative mediator integrating diverse resources to generate solutions in academic, professional, and interpersonal contexts. Trust in AI, however, remained conditional, reflecting recognition of its limited autonomy and restricted reliability in expert domains. Emotionally and morally, AI was consistently framed as a non-agentive, emotionally neutral entity. Its affective expressions were interpreted as imitations rather than authentic emotions, and interactions were evaluated through human-centered criteria of utility, risk, and safety. This study provides phenomenological insights into human–AI interaction, showing that humans primarily view AI as an instrumental tool bounded by efficiency and safety. Ultimately, it raises a fundamental question: what constitutes genuine interaction between humans and AI, establishing a basis for future ethical discourse and development.

Abstract
1. 서론
2. 연구 대상과 방법
3. 연구 결과
    3.1. 분석적 지능을 갖춘 소극적 문제해결자
    3.2. 사용자 환경에 국한된 수동적 존재
    3.3. 인간 중심적 목적에 종속된 형식적 감정의 존재
    3.4. 도덕적이어야 하는 인턴 정도의 비독립체
4. 결론
참고문헌

• 김지영(Ph.D. Student, Department of Korean Language and Literature, Yonsei University) | Jiyeong Kim