Abstract:

As artificial intelligence becomes part of everyday school practice, many teachers are interested in using it, but far fewer feel confident about how to use it meaningfully in their own subject. General discussion about AI in education is growing, yet subject teachers still need practical answers: What should students learn about AI in my subject? Which activities make sense? How can AI support learning without becoming just a shortcut for convenience? These questions are especially important in schools where access to AI tools is uneven, and teachers’ experience with them varies widely.

This article focuses on the development and evaluation of subject-specific AI learning materials for grades 7–12 in Estonia in mathematics, music, literature, English, Estonian, history, and civics. The materials are designed for all Estonian teachers and currently include around 35 examples of how teachers can use AI in their work and about 100 examples of how AI can be used with students. The materials are being evaluated as part of a broader teacher support initiative linked to short national AI training courses and the year-long AI-Leap professional development program, in which 250 teachers explore both the philosophical and practical dimensions of AI in education. Most teachers are familiar with tools such as ChatGPT and Gemini, and to a lesser extent, Copilot, Grok, and others. However, knowing the tools does not automatically mean knowing how to use them well for learning.

A key challenge is that even subject experts often find it difficult to design good AI-supported learning activities. Interest is high, but confidence is often low. Many experts struggle to turn general enthusiasm into subject-appropriate teaching ideas. There is also a fear of peer criticism: teachers may hesitate to suggest examples because they worry colleagues will see them as weak, simplistic, or technically uninformed. As a result, development may stall not because teachers reject AI, but because the threshold for creating good examples feels too high.

Our experience shows that when teachers are simply told to “use AI,” the results are often weak. Activities may become vague, tool-driven, or too focused on efficiency. In such cases, AI is easily treated either as a substitute teacher or as a tool that does the work for students. This does not support deep learning and may weaken understanding, learner agency, and academic honesty. For this reason, the materials are not built around AI as a general productivity tool. Instead, they are designed around a specific learning activity, a clear subject goal, and the development of learner agency. AI is framed as a learning partner that can support questioning, reflection, drafting, comparison, feedback, rehearsal, and idea generation, but not replace the teacher or the learner’s own thinking.

The materials combine two levels: a shared foundation across subjects and subject-specific sections. The common part introduces what AI means in education, what learning with AI involves, and how AI can be used responsibly. The subject-specific parts address teachers’ actual concerns and provide practical examples of meaningful classroom use. We argue that teachers need more than access to AI tools. They need clear subject-specific examples that build confidence and connect AI use to real learning goals.

Keywords:

Artificial Intelligence in Education, Subject-Specific Materials, Teacher Support, Learning Design, AI Literacy, Learner Agency, Professional Development.