AI and the future of Learning: Design guidelines for group work and feedback processes in primary education in Switzerland
This thesis explores how artificial intelligence can be responsibly integrated into Swiss primary education to support collaborative group learning and feedback processes. It highlights both the potential and the boundaries of AI in early learning environments.

Topic
This thesis explores how artificial intelligence can be responsibly integrated into Swiss primary education to support collaborative group learning and feedback processes. Combining educational theory with insights from primary school teachers, it highlights both the potential and the boundaries of AI in early learning environments. The study delivers concrete, child-centered design guidelines that bridge the gap between technological innovation and pedagogical reality, advocating for transparent, age-appropriate AI systems that respect human relationships, maintain teacher agency and promote learner autonomy.
Relevance
With ongoing technological advancements worldwide, the role of AI in Swiss primary education is expected to grow. Early academic engagement offers an opportunity to inform these discussions in a targeted, practice-oriented way that helps educators and policymakers shape this development and support responsible integration. The goal is not to replace pedagogical judgment, but to complement it by ensuring that AI strengthens key competencies such as collaboration, self-regulation and problem-solving, rather than undermining them. This makes the thesis highly relevant for practitioners who are navigating the digital transformation in schools.
Results
The results of this thesis show that Swiss primary school teachers are cautiously open to the use of AI, provided it is developmentally appropriate, transparent and remains under pedagogical control. AI is generally welcomed as a supportive tool for tasks such as planning, organizing, observation or suggesting group compositions, particularly in upper primary levels. However, teachers across all levels reject autonomous decision-making and emotionally nuanced interactions. In early childhood education, the importance of human relationships, sensory experiences and experiential learning is especially emphasized. The study provides concrete, child-centered design guidelines for the responsible AI integration aligned with classroom realities.
Implications for practitioners
- Introduce AI literacy starting from grade 4
AI education should begin only once students have developed basic critical thinking skills. In earlier grades, the focus should remain on physical development, language acquisition and foundational skills. - Direct AI feedback to students is only appropriate in narrowly defined practice tasks
Automated feedback may be used in factual exercises such as vocabulary or arithmetic, where it is important to prevent students from internalizing incorrect information that could negatively impact the learning process. However, teachers do not support its use in open-ended, collaborative or social learning situations. - AI-generated reflection prompts should follow Socratic principles
Prompts must stimulate independent thinking through open-ended questions without offering predefined answers. They should foster metacognition while keeping pedagogical control with the teacher. - AI may help visualize group dynamics, but interpretation remains with the teacher
Systems can unobtrusively track participation patterns, but these data must be used as supportive input only and never replace professional judgment.
Methods
This thesis used a qualitative research design within a Design Science Research (DSR) framework to explore how AI can meaningfully support group learning and feedback processes in Swiss primary education. Eleven semi-structured interviews were conducted with teachers from all three primary school levels: kindergarten, lower primary (grades 1–3) and upper primary (grades 4–6). The study aimed to capture their perspectives on AI’s potential, limitations and implementation challenges. The interview data were thematically analyzed to identify key insights. Based on the findings, actionable design recommendations were developed and validated through participant feedback to ensure their relevance for practice and policy.