Interplay effects of AI feedback and individual differences on learners’ emotional responses

Computer and AI feedback practice has received increasing attention in educational studies. Nevertheless, a line of research remains underexplored regarding the interplay between AI feedback types and the learners’ emotional responses along with individual difference. To address this gap, this study collected data over two weeks of 396 feedback responses provided by generative AI Chatbot from 66 engineering students in a university on their scientific writing reports. Survey data was triangulated with the learners’ emotional responses towards the combination of three AI feedback types on two report forms along with learner’s individual difference of writing proficiency and feedback literacy. Hierarchical regression analysis supported all the hypotheses that Model 1 of individual difference, Model 2 of newly-added AI feedback type and Model 3 of newly-added report type all significantly predicted the learners’ positive emotional responses, while all the three models failed to predict the negative emotional responses. In addition, paired T-test yielded statistically higher scores in form-focused type of negative “State of bored” within evaluative type and “High arousal negative state” within suggestive type. Likewise, paired T-test results generated higher scores in evaluative feedback of positive “Positively surprising state” and negative “State of bored” within form-focused type. The implications of the findings are discussed on the development of learners’ AI feedback literacy.