Mobile Language App Learners' Self-Efficacy Increases After Using Generative AI

Audrey Kittredge ^* Elise Hopman Ben Reuveni Danielle Dionne Cassie Freeman Xiangying Jiang

• Duolingo, Pittsburgh, United States

Introduction: Although generative artificial intelligence (AI) is ubiquitous, there is little research on how it supports self-efficacy (learners’ belief that they can perform at a particular level on a specific task). The purpose of these studies was to investigate self-efficacy development in a generative AI-based language learning experience. Methods: In two studies, learners (N = 385) of French/Spanish used AI-based features offering conversation practice and on-demand explanations in a mobile app (Duolingo) for one month. Before and after using the features, learners reported their self-efficacy and other perceptions.Results: In Study 1, learners who had already used the features felt significantly more prepared to use French/Spanish in real-life situations after one month, as did learners in Study 2 who used the features for the first time. Learners in Study 2 also felt significantly more prepared to share their opinions and navigate a city, and reported significantly higher self-efficacy for speaking and understanding grammar and mistakes. Across studies, the majority of learners agreed that the AI-based features effectively supported learning, and reported using their learning outside the app. Discussion: These results provide the first evidence of enhanced language learning self-efficacy after use of generative AI, building on findings from classroom interventions.