Inclusive AI in diverse linguistic contexts: A phenomenological exploration of user experiences with ChatGPT

Abstract

Purpose: This study investigates how using a first or second language influences user perceptions of ChatGPT. It aims to better understand how multilingual contexts shape AI-human interactions. Design/methodology/approach: After participants interacted with ChatGPT in both their native language (Turkish) and second language (English), semi-structured interviews were conducted. We conducted a phenomenological analysis that examined emotional engagement, human likeness, cultural fluency, and perceived intelligence in both languages. Findings: Participants with lower English proficiency attributed more humanlike qualities to ChatGPT in their native language, Turkish. They also reported stronger emotional engagement in their native language. However, most participants perceived English interactions as more efficient and intelligent, and concerns arose over ChatGPT’s cultural fluency. Research limitations/implications: Balancing emotional connection with linguistic and cultural accuracy is tricky for non-native speakers of English. These results show the importance of language and cultural context in shaping perceptions of AI The study was conducted with 33 participants, and the findings are not generalizable to the whole population. Practical implications: The study offers valuable insights that help researchers and professionals achieve a broader understanding of AI-human interaction across languages. Therefore, AI systems must improve linguistic proficiency and cultural fluency to be more inclusive and humanlike. Originality/value: This study explores the experiences of non-native English speakers with AI.