Student–AI research mentor collaboration: Initial findings from an introductory research methods course

Creative Commons License

Saykılı A.

International Conferences E-Learning and Digital Learning 2025 and Sustainability, Technology and Education 2026 , Valencia, İspanya, 24 - 27 Temmuz 2026, ss.139-146, (Tam Metin Bildiri)

  • Yayın Türü: Bildiri / Tam Metin Bildiri
  • Basıldığı Şehir: Valencia
  • Basıldığı Ülke: İspanya
  • Sayfa Sayıları: ss.139-146
  • Açık Arşiv Koleksiyonu: AVESİS Açık Erişim Koleksiyonu
  • Anadolu Üniversitesi Adresli: Evet

Özet

This study reports initial findings on student–AI research mentor collaboration during topic and research question (RQ) development in an undergraduate level research methods course. The focal workflow included two stages: teams first drafted a topic and primary RQ without AI support (Assignment 4.1), then individuals consulted an AI mentor and returned to their teams to reconcile recommendations through an explicit accept/adapt/reject process (Assignment 4.2). Following Assignment 4.2, 58 of 74 students (78.4%) completed a survey including 12 Likert-type items (1–5) and three open-ended prompts. Quantitative results indicated strong perceived learning and refinement value, with the highest endorsement for improved understanding of what constitutes a good research question (Item 10, M = 4.43) and high overall recommendation intentions (Item 11, M = 4.52). Feedback clarity was rated favorably (Item 4, M = 4.19), while perceived project-specificity and comfort disagreeing were comparatively lower and more variable (Items 5–8; lowest for Item 8, M = 3.86). Thematic analysis showed that students primarily used the AI mentor to narrow overly broad topics, increase specificity (participants/context/variables), and refine wording; disagreement episodes centered on generic feedback, over-specification demands, and method-type misclassification, typically resolved through feasibility checks, instructor-aligned criteria, and team deliberation. Findings suggest that pairing an explicit checklist with AI mentoring can support calibrated uptake and learner agency in early research design tasks.