Academic emotions and perceived learning outcomes in GenAI-supported learning: the serial mediating roles of AI acceptance and learning behaviors

With the widespread application of generative artificial intelligence (GenAI) in higher education, academic emotions play a vital role in students' learning processes. This study aims to examine the associations between positive and negative academic emotions and college students' perceived learning outcomes, and analyze the serial mediating roles of AI acceptance and learning behaviors within GenAI-supported learning environments.

A total of 526 college students with GenAI learning experience were recruited via convenience sampling. All participants completed four standardized scales measuring academic emotions, AI acceptance, learning behaviors and perceived learning outcomes. Descriptive statistics, correlation analysis, confirmatory factor analysis and bias-corrected Bootstrap mediation tests were conducted using SPSS 26.0 and AMOS 27.0.

All variables were significantly correlated (p < 0.01). Positive academic emotions exerted a significant direct association with perceived learning outcomes (β = 0.176), whereas the direct association of negative academic emotions was non-significant (β = -0.098, p = 0.049, 95% CI [-0.205, 0.007]). AI acceptance and learning behaviors had significant independent mediating effects. The serial mediation effect value was 0.020 for positive emotions and -0.037 for negative emotions, with all 95% confidence intervals excluding zero. Indirect pathways dominated the influence of negative academic emotions.

This study identifies a pattern of associations consistent with the hypothesized emotion-cognition-behavior-outcome pathway in GenAI-supported learning. The findings offer empirical evidence for developing targeted academic-emotion-informed intervention strategies, improving students' AI acceptance and learning engagement, and promoting the high-quality development of intelligent education.