Generative AI has been a growing topic in classrooms for quite some time now – unsurprisingly, students have long since figured out how to use tools such as ChatGPT to write their homework or generate summaries to get through their research faster.
What is new, however, are the demands on teachers to use generative AI themselves, for example to create teaching materials or assess homework.
Wikimedia CH’s Innovation Manager Shana Marinitsch was invited to a conference for teachers at KS Stadelhofen to give an assessment of what current models can and cannot do, what teachers can learn from the source work of Wikipedians for their own teaching, and how they can teach good, long-term useful skills in the classroom without creating a cycle of generative content that is in turn evaluated by generative models.
A highly regarded study was recently published on this topic, which was incorporated into the presentation and shows that, unlike search engines and traditional source work, the use of LLMs has a significant impact on cognitive skills in essay writing: ‘Your Brain on ChatGPT’ by MIT (open access via arxiv.org – [2506.08872] Your Brain on ChatGPT: Accumulation of Cognitive Debt when Using an AI Assistant for Essay Writing Task)
A particularly exciting outcome of the conference was that teachers were open, self-critical, but also pragmatic in their re-evaluation of the rules for Matura papers in particular, but also for homework in general, in order to strengthen important skills among students such as methodological work, media literacy and critical thinking.
We are pleased about the exchange and solidarity with experts themselves, which enables us to make the values surrounding open knowledge and the source work of the Wikiverse accessible to everyone in a way that is close to real life.
The presentation slides themselves are, of course, available under a CC BY-SA licence on Wiki Commons.
