Why do students discontinue learning with ChatGPT? The role of negative information in reducing motivation and learning discontinuation

As conversational AI tools like ChatGPT become increasingly integrated into education, concerns have emerged about how negative information (linguistic cues emphasizing uncertainty, difficulty, or risk) may influence adolescents’ trust and learning behaviors. This study examined associations between negative ChatGPT responses, adolescents’ AI distrust, and learning-related outcomes, including motivation and discontinuation. Among 250 adolescents aged 12–13 who completed a 10-minute ChatGPT-based learning activity on inductive and deductive reasoning, analyses of 195 participants indicated that greater frequency of and initial exposure to negative information were associated with higher AI distrust, which was related to lower learning motivation and greater learning discontinuation. Moreover, dissatisfaction with ChatGPT’s responses moderated these relationships, such that associations between AI distrust and lower learning motivation and greater learning discontinuation were stronger among more dissatisfied adolescents. These findings underscore the importance of how negative AI-generated information is communicated, given its potential implications for adolescents’ trust, motivation, and continued learning engagement.