AI is becoming increasingly embedded in educational practices (Zou et al., 2025). While current research has largely focused on the general promises and challenges of AI in education (Bond et al., 2024; Vieriu & Petrea, 2025), less attention has been paid to how students, especially younger ones, understand their own roles within these AI-enhanced learning environments. Yet, students play a critical role in how pedagogical relations play out, even as AI increasingly shapes these relations as well as modes of participation (Perrotta, 2024).

Drawing on a sociomaterial perspective, we conceptualize AI-enhanced learning environments as assemblages of human actors, technologies, institutional structures, and practices (Fenwick, 2015; Fenwick & Landri, 2012; Sørensen, 2001). From this perspektive, we understand students’ imaginaries of AI as ways of making sense of their roles in data-driven education, including how participation and responsibility are configured within such assemblages. Consequently, we ask: How do secondary school students perceive and negotiate their roles within AI-enhanced learning environments? And how can Computational Empowerment (CE) support critical reflection on these sociomaterial assemblages?

This concise paper draws on a series of workshops with 75 students across three secondary schools and presents preliminary findings from an ongoing study. Building on CE (Dindler et al., 2020, 2023) as a participatory design approach, students were invited to imagine and prototype their AI-enhanced ideal learning platforms. The workshops were intended to elicit students’ imaginaries of AI integration and to create a space in which they could question and reconfigure their roles within these data-driven learning environments.

Our preliminary findings suggest that students imagine AI mainly as a personalized tutor that supports individual performance. They expect it to explain content, generate exercises to assist with exam preparation, and serve as a central hub for accessing information. Rather than envisioning AI as transforming education, they position it as reinforcing existing school logics of efficiency, task completion, and individual achievement. Students imagine it as an omniscient mediator of knowledge and themselves primarily as dependent users rather than active participants in learning processes. They also attribute social and regulatory functions to AI, such as moderating communication and enforcing rules on their platform. Overall, our findings suggest that students imagine AI not as a tool but as an active participant in the educational assemblage, one that tends to reproduce existing pedagogical and institutional practices. This raises further questions about how responsibility is imagined when AI assumes functions otherwise associated with teachers, and what this means for students’ sense of agency and participation. These questions will be discussed in the presentation and explored in the final paper.

Methodologically, CE enabled students to reflect on tensions within AI-enhanced learning environments and revealed how their imaginaries of AI often reproduce dominant school logics. At the same time, CE proved valuable for making these tensions visible. It opened up discussion about the pedagogical and relational dimensions of AI-enhanced education, highlighting its value as a method for researching these environments as sociomaterial assemblages.