Learning to Learn with AI – Guiding First-Semester Students through Reflection, Feedback, and Student-Centered Learning

Implementation of the Course

Course context 
World Food System (Welternährungssystem) is a mandatory first-semester lecture series for students in Agricultural Sciences (D-USYS), Food Science and Nutrition (D-HEST), and and a very popular elective course in Environmental Sciences (D-USYS). Taught by multiple research groups, the course introduces students to the complexity and interconnectedness of global food production systems, including environmental, social, and economic dimensions. It provides foundational disciplinary knowledge while fostering systems thinking across fields of study.
Within this context, the accompanying learning element was redesigned as a direct follow-up to the POT-AI Innovedum project. The aim was to embed the critical and responsible use of AI tools into students’ early academic practice and to guide them in integrating AI meaningfully into their learning processes.

Design of the Learning Task  
The redesigned learning element consists of three structured tasks distributed across the semester. Using content from the World Food System lectures, students are guided step-by-step in learning how to use AI tools for academic purposes, such as drafting texts, identifying and validating literature, and critically evaluating AI-generated outputs. A strong emphasis is placed on academic integrity and compliance with university guidelines.
The tasks are introduced in the lecture hall and subsequently carried out online via Moodle and the reflection tool Rflect. Throughout the semester, students engage in guided reflection and self-assessment cycles. They define personal learning goals, evaluate their strategies and AI use after each task, and identify areas for improvement. Their activities are monitored and guided by the team of student teaching assistants (STAs). Students can ask questions through Moodle forums and are encouraged to attend one on-site tutoring session (KInspiration Bazar in HS25).

Role of the Student Teaching Assistants
The STAs play a central role as co-designers and facilitators of the initiative. They actively contribute ideas, refine task instructions, and help monitor and respond to student reflections. For the 2025 iteration, STAs were particularly engaged in developing the reflection prompts and shaping the self-assessment process. Their proximity to first-year students enabled them to design formats that were accessible, motivating, and relevant.

Self-Reflection Process with Rflect 
At the beginning of the semester, students completed a self-assessment aligned with defined AI-related competences and could compare their results with aggregated class data. Before each task, they reflected on individual learning objectives; afterward, structured prompts guided them in evaluating their approach, AI use, and outcomes.
The self‑reflection prompts also invited students to indicate any questions or difficulties, giving the team deeper insight into students’ learning processes and highlighting challenges that might otherwise have remained invisible.
Rflect’s AI-supported features automatically summarized responses and flagged entries requiring attention, enabling targeted feedback and making the approach scalable.
Through recurring reflection cycles, students deepened their understanding of the course content, strengthened metacognitive skills, and developed sustainable self‑regulated learning habits. By the end of the semester, students repeated the competence self-assessment, showing measurable development in targeted areas.

KInspiration Bazar 
A highlight of the semester was the “KInspiration Bazar,” fully designed and organized by the STAs. Inspired by the Coding Bazar format (developed by D-INFK lecturers Lukas Fässler and Markus Dahinden), the event offered interactive stations where students explored AI tools, discussed ethical and environmental implications, identified AI-generated content, and experimented with tools for their own learning. A discussion on AI-generated peer feedback sparked a deeper conversation about the value of human feedback in academic learning.
Participation and engagement were remarkably high. The appealing format and strong communication by the STAs, together with the ongoing reflection cycles, contributed to active involvement and a lively and motivating learning atmosphere.

Scalability and lessons learned 
The pilot demonstrates that combining structured reflection, targeted AI competence development, and active STA involvement can significantly enrich first-year learning experiences. Guided reflection cycles, supported by scalable digital tools, foster student ownership, metacognitive awareness, and critical AI literacy. High participation rates and observable competence gains indicate strong potential for transferring and scaling this model to other courses and, ultimately, to the curricular level.

Motivation, Project Mission, Vision Statement 

Our motivation is to respond to the growing need for students to navigate AI-rich learning environments critically, responsibly, and self-directedly from the very beginning of their studies. This pilot aims to redesign a first-semester learning element so that structured self-assessment and guided reflection become integral parts of learning how to use AI in academic contexts. Our mission is to create a student-centered environment in which reflection is not an add-on, but a systematic practice that strengthens metacognition, ownership of learning, and sustainable study habits. Beyond the course level, this initiative serves as an experiential foundation for the ongoing curriculum revision at D-USYS (Studiengangsinitiative²). It provides concrete insights into how reflection and self-assessment can be meaningfully and structurally embedded across the study program. Through close collaboration with STAs and the integration of student‑centred learning principles, this approach aspires to shape a culture of curiosity, responsibility, and continuous growth that accompanies students throughout their learning journey at ETH and professional life thereafter.

Innovative Elements 

This initiative stands out through:

• Integration of AI‑competence development into a mandatory first‑year course
• Semester‑long reflective learning architecture using Rflect
• Scalable, AI‑supported analysis of student reflections
• Student teaching assistants as co‑designers and facilitators
• Experiential formats such as the KInspiration Bazar

Effects on Student Learning

The pilot shows clear positive effects on student learning and engagement. Around 95% of participating students completed all reflection cycles and self-assessments on time, indicating a high level of commitment. While only seven students reported that they did not particularly enjoy the reflection elements, they nevertheless engaged seriously with them and demonstrated meaningful qualitative progress in their responses. On average, students invested between 11 and 87 minutes across six reflection cycles and two self-assessments. Reflection prompts encouraged students to critically evaluate their learning strategies, question the reliability of AI tools, and articulate challenges. Comparative pre- and post-self-assessments indicate measurable gains in targeted AI-related competencies, especially in the critical and responsible use of AI.

Acknowledgement

The S4S learning element presented here is constantly being developed. The following members of the team also contributed significantly to the original concept and design through their hard work and valuable input: Sara Ammann, Anna Klopfer, Mikayla Hug, Doreen Haueter, Levi Brucker, and Nicolas Dähler