BERLIN, June 13, 2026 — During the summer semester of 2026, students and lecturers from Humboldt-Universität zu Berlin and the University of Oslo came together for the Circle U. COIL course AI-Supported Lesson Development.
Combining in-person sessions in Berlin with international online exchange, the course explored how generative artificial intelligence can support lesson planning while encouraging critical reflection on its pedagogical, ethical, and technological implications. Across ten sessions from April to June, students experimented with AI tools, developed prompting strategies for educational practice, analyzed existing AI applications for schools, and discussed issues such as data protection, algorithmic bias, copyright, and technological solutionism.
A central question guided the course: How can AI tools be meaningfully and critically integrated into lesson preparation?
Students worked with Humboldt-Universität's centrally hosted AI platform as well as locally installed applications such as Ollama, ChatboxAI, and AnythingLLM. Through hands-on exercises, they developed systematic prompting techniques and explored how generative AI can support lesson planning, differentiation, assessment, feedback, and the creation of teaching materials.
Students also analyzed education-focused AI services such as MagicSchool AI, fobizz, and fellowfish, examining not only what these tools can do, but also the assumptions and limitations behind their design. These activities contributed to the development of a shared prompt library with strategies and examples for different teaching situations.
Two joint online sessions in May brought together students and lecturers from Berlin and Oslo. Conducted in English, the sessions provided opportunities to compare German and Norwegian perspectives on AI in education and to discuss responsible AI use, GDPR, AI business models, AI-detection tools, and technological solutionism.
Insights into current AI practices in Norwegian schools further demonstrated how educational and national contexts influence the implementation of new technologies.
While the later production phase was primarily carried forward by the Berlin students, the international sessions provided important perspectives that informed the course's practical and critical work.
During the second half of the course, students transformed their experiences into educational podcasts on the opportunities, limitations, and ethical implications of generative AI in schools.
The podcast projects connected practical experimentation with broader questions of transparency, bias, data protection, and pedagogical responsibility. Digital tools including Moodle, Etherpad, OnlyOffice, Padlet, and Zoom supported collaboration and exchange throughout the course.
The aim was not simply to learn how to use AI more efficiently. Students were encouraged to consider when AI can genuinely support teaching, where its limitations lie, and which professional and pedagogical responsibilities should remain with educators.
The course concluded on 13 June with a full-day online Final Fair. Students presented their podcasts and reflected on their experiences with AI-supported lesson development and critical prompting.
The programme was complemented by expert contributions. Danny Walther presented approaches to AI-supported literature and data research, with particular attention to information literacy and the identification of hallucinated references. Julie Lüpkes (Carl von Ossietzky Universität Oldenburg) contributed findings from her ethnographic research on an EdTech startup developing an AI-based assessment tool, highlighting the influence of market dynamics, organisational practices, and pedagogical assumptions on educational technology.
The COIL course showed that AI literacy in teacher education goes beyond learning how to operate generative AI tools. It requires practical experimentation combined with critical reflection on privacy, bias, transparency, and pedagogical responsibility. By bringing together technical, educational, ethical, and international perspectives, AI-Supported Lesson Development created a foundation for further integrating responsible and practice-oriented AI literacy into teacher education.
Funded by the European Union. Views and opinions expressed are however those of the author(s) only and do not necessarily reflect those of the European Union or the European Education and Culture Executive Agency (EACEA). Neither the European Union nor EACEA can be held responsible for them.
© 2025 AI2PI Academy. All rights reserved. Project illustrations by Cipuva
