This study was conducted to examine the effects of feedback provided by a generative AI chatbot trained by the researcher and feedback given by a teacher on students’ achievement and retention in informative text writing. In this context, a convergent parallel mixed-methods approach was adopted. The research was conducted with two groups at the fifth-grade middle school level, comprising experimental and control groups. For eight weeks, the chatbot provided feedback on the texts of students in the experimental group through the researcher, while the teacher provided feedback on the texts of students in the control group. To measure students’ informative text writing performance, the Informative Text Writing Performance Assessment Scale developed by Hamzadayı and Dölek (2022) was utilized. Findings indicated that feedback provided by both the AI chatbot and the teacher contributed to improvements in students’ informative text writing performance. However, it was determined that such improvement persisted significantly only within the teacher-feedback group. Based on this finding, it was concluded that while chatbot feedback was as effective as teacher feedback in improving writing performance, it did not exhibit comparable long-term retention effects. When comparing the content of feedback provided, it was observed that the chatbot’s feedback was quantitatively more extensive than that of the teacher. Conversely, teacher feedback was superior regarding affective components and question-oriented feedback. Consequently, generative AI tools hold significant potential for reducing teachers’ cognitive workload in writing instruction by providing supplementary and innovative feedback methods.
Baz et al. (Mon,) studied this question.