Integrating Generative AI Tools by Students in the Preparation of Final Qualification Research Papers in Foreign Language Teaching Methodology

Modern generative artificial intelligence (AI) tools can address many research objectives faced by students when conducting studies and writing final qualification research papers (FQRPs). However, AI’s propensity for hallucinations, difficulties in prompt engineering, and students’ lack of systematic understanding of the legal use of AI in education and research lead to fragmented and task-specific AI adoption. This study aims to assess the extent to which students leverage generative AI tools to address research objectives when preparing FQRPs in foreign language teaching methodology. The research employs multiple methods: (a) surveying, (b) statistical analysis of survey results, and (c) expert evaluation. Materials included review and research articles from Russian and international journals indexed in Scopus, Web of Science, and the Russian Higher Attestation Commission (VAK) list for pedagogical sciences. The empirical study was conducted at Derzhavin Tambov State University, involving 4th-year students working on their FQRPs: an experimental group (EG, n = 48) majoring in Pedagogical Education (English Language) and Linguistics (Theory and Methods of Teaching Foreign Languages and Cultures), and a control group (CG, n = 48) majoring in Linguistics (Translation Studies) and Philology (Foreign Philology). Data were collected via an open-ended online questionnaire, where students indicated which research tasks they used generative AI for. Responses from diagnostic and control stages were coded and analyzed using Student’s t-test. The study revealed that with official permission and regulated use of generative AI, students employed it for the following research tasks in FQRP preparation: drafting the FQRP outline (t = 2.59, p < 0.05), sourcing scientific materials (t = 3.07, p < 0.05), formulating arguments (t = 2.83, p < 0.05), developing teaching/language materials (t = 6.57, p < 0.05), editing (t = 2.83, p < 0.05), and writing FQP text (t = 1.77, p < 0.05). No statistically significant differences were found between EG and CG in using AI for data visualization (t = 1.43, p > 0.05) or reference list formatting (t = 1.43, p > 0.05), suggesting that students who chose AI for these tasks were already proficient in its application, while others abstained. The study highlights five key issues requiring broader discussion: (a) the rapid proliferation of AI and its capabilities/limitations; (b) academic ethics and AI-assisted plagiarism; (c) the role of supervisors in student FQRP writing; (d) the methodological and linguistic research potential of AI-based tools; and (e) the need for new forms of final assessments feasible only for humans. The authors declare no conflict of interests.

Keywords

Authors

References