Trust and Authority in AI-First Language Learning: The Case of Duolingo

Abstract Book of the 12th International Conference on Social Sciences, Humanities and Education

Year: 2026

[PDF]

Trust and Authority in AI-First Language Learning: The Case of Duolingo

Ashley Leung

ABSTRACT:

In 2025, Duolingo shifted to an “AI-first” model, prompting concerns about replacing human instruction with automated content. This study examines how the transition influences learner trust, perceptions of instructional authority, and engagement in language learning. Using a mixed-methods approach, university students in China completed an online survey measuring trust, perceived usefulness, satisfaction, and awareness of the AI-first shift. Follow-up semi-structured interviews explored participants’ experiences, emotional engagement, and interpretations of Duolingo as a teacher, tool, platform, or AI system. Findings are expected to show that trust varies by task type, with higher confidence in routine exercises and lower confidence in culturally nuanced or open-ended learning activities. Skepticism toward AI and awareness of public backlash may lower perceived credibility and accountability. Interview data are expected to reveal mixed understandings of instructor identity, emphasizing the role of perceived human presence in engagement. This research contributes to human-AI trust scholarship and informs AI-powered language-learning platform design, highlighting the importance of transparency, learner-centered communication, and balancing automation with human-like pedagogical support.

Keywords: AI; Digital Education; Duolingo; Human-AI Trust; Language Learning