Peer Collaboration and Generative AI Interaction: Assessing Student Agency in Project-Based Programming
16th International Conference on The Future of Education, FOE 2026, Florence, İtalya, 18 - 19 Haziran 2026, (Tam Metin Bildiri)
- Yayın Türü: Bildiri / Tam Metin Bildiri
- Basıldığı Şehir: Florence
- Basıldığı Ülke: İtalya
- Anahtar Kelimeler: Collective Agency, Generative AI, Human-AI Co-agency, Programming Education, Project-Based Learning, Student Agency
- Ankara Üniversitesi Adresli: Evet
Özet
Generative artificial intelligence (GenAI) and large language models are increasingly integrated into project development processes; however, limited empirical evidence explains how students preserve agency when working concurrently with peers and AI systems. This qualitative study examined how associate degree students positioned GenAI during a project-based programming course and how individual and collective agency emerged in this context. The study group consisted of 42 students enrolled in a Computer Programming Program at a vocational school in Türkiye in 2026. The students developed software projects in 14 collaborative groups over a 14-week Professional Project course. The data were collected through an open-ended questionnaire addressing students' experiences of AI use, decision-making, cognitive support, and collaboration. The data were analyzed through directed content analysis, informed by the concepts of human-AI co-agency, student agency, cognitive scaffolding, and collaborative learning. The findings indicated that students predominantly positioned GenAI as a teammate or mentor rather than as a merely technical tool. They associated AI support mainly with scaffolding, reflective dialogue, and cognitive load reduction. Although AI was used extensively, the students reported that final decisions were generally made through group discussion, critical questioning, verification, and independent judgment. Most participants preferred a hybrid collaboration model that combined AI support with peer interaction. These findings suggest that, when embedded in a pedagogically structured project-based learning environment, GenAI can support problem-solving / project-based programming / project-based learning and collaborative work without necessarily diminishing students' agency. The study offers implications for designing human-centered AI-supported learning environments that maintain learners' responsibility, critical engagement, and collective decision-making.