To address the ever-growing global challenges, schools aim to educate students with systems thinking skills, a strong foundation in natural sciences, the ability to think creatively, recognize interdisciplinary connections, and collaboratively apply these insights to problem-solving. In many countries, a significant issue is the lack of students' interest in learning natural sciences. Various studies have shown that interdisciplinary teaching increases students' engagement and motivation, as well as their understanding of concepts and skills in natural sciences. An interdisciplinary science curriculum enhances students' understanding of scientific concepts and their ability to apply these concepts in real life. Interdisciplinary teaching can be defined and implemented in various ways. There have been several models created to describe interdisciplinary teaching. The most comprehensive model is considered to be the one presented by Labudde in 2005, but the author himself has already left one dimension of the model empty for future developments and enhancements. To understand the complexity and recognize the shortcomings of interdisciplinary teaching, a new model of interdisciplinary teaching is needed. For example, none of the existing models consider factors such as national exams and assessments, the impact of resources or institutional support. Our aim was to create a comprehensive model that allows us to evaluate all aspects of interdisciplinary teaching. This aim led to the research question: Can AI effectively categorize different aspects of interdisciplinary teaching into new model and how similar is the AI categorization to the categorizations of the human experts?

Based on previous models and studies a new model of interdisciplinary teaching was developed. The new model consists of three main dimensions: organization of learning, preparation and learning process. Each dimension is further divided into sub-dimensions. For example, the organization of learning consists of three sub-dimensions: management; curriculum and schedule; exams and assessments. Each sub-dimension has several sub-categories.

In spring 2024 a study with 53 Estonian science high school teachers was conducted to examine the understanding of interdisciplinary teaching, its importance and the methods used. As part of the study, respondents were asked to list up to 10 keywords they consider important for interdisciplinary teaching. Altogether teachers listed 446 keywords. To categorize the responses into the new model, the collected keywords were given to three long-term science teacher experts, who have conducted research in interdisciplinary teaching. After that, the AI application ChatGPT was given the same task: to categorize the responses into the new model. The expert categorizations were compared with the ChatGPT categorizations. The results showed an overall 18% overlap between ChatGPT and the experts' categorizations. However, ChatGPT achieved over 50% agreement with two of the experts, demonstrating its potential in categorizing interdisciplinary teaching concepts. These findings highlight AI’s potential as a complementary tool in educational research.

Keywords: Science education, interdisciplinary teaching, AI, validation, expert ratings.