Does AI Guidance make a difference? a quasi-experimental study on the effects of NotebookLM on academic achievement, test anxiety, and self-efficacy
Education and Information Technologies, 2026 (SSCI, Scopus)
- Yayın Türü: Makale / Tam Makale
- Basım Tarihi: 2026
- Doi Numarası: 10.1007/s10639-026-14146-2
- Dergi Adı: Education and Information Technologies
- Derginin Tarandığı İndeksler: Social Sciences Citation Index (SSCI), Scopus, IBZ Online, Agricultural & Environmental Science Database, EBSCO Education Source, Education Abstracts, Educational research abstracts (ERA), ERIC (Education Resources Information Center), INSPEC, Social Science Premium Collection (ProQuest), Education Collection (ProQuest), Education Source Ultimate (EBSCO)
- Anahtar Kelimeler: Academic achievement, Academic self-efficacy, AI-assisted learning, Generative artificial intelligence, NotebookLM, Test anxiety
- Akdeniz Üniversitesi Adresli: Evet
Özet
This study examined the effects of NotebookLM-assisted learning on university students’ academic achievement, test anxiety, and academic self-efficacy. Using a pretest-posttest control group quasi-experimental design, 118 undergraduate students enrolled in a Scientific Research Methods course were randomly assigned to three groups: NotebookLM + teacher guidance, independent NotebookLM use, and traditional instruction. After the six-week intervention, data were analyzed using 3 × 2 mixed ANOVA with Bonferroni post-hoc comparisons. Results revealed statistically significant time × group interactions for all three dependent variables (academic achievement: p <.001, η²p =.46; test anxiety: p <.001, η²p =.41; academic self-efficacy: p <.001, η²p =.59). For academic achievement, the guided group significantly outperformed both the independent (p =.036) and the control groups (p <.001), while the independent use group also significantly outperformed the control group (p =.002). For test anxiety and academic self-efficacy, both experimental groups improved significantly more than the control group (ps < 0.001-0.010), with no difference between them (both ps = 1.000). These results indicate that the effects on anxiety and self-efficacy operated at the instrumental level, meaning they were driven by tool use alone regardless of guidance, whereas the effect on academic achievement required pedagogical support. The findings provide experimental evidence that AI should be positioned not as a replacement for the teacher, but as a complementary tool integrated with pedagogical scaffolding.