This chapter focuses on generating STEM learning tasks of varying difficulty according c Taxonomy for physics and chemistry at secondary schools. STEM education integrates science, technology, engineering, and mathematics, providing a holistic approach to teaching that goes beyond the traditional boundaries of individual subjects. Based on the revised Bloom Taxonomy, Tollingerová’s Learning Task Taxonomy offers a structured framework for creating tasks that develop students’ various cognitive processes. Generative Artificial Intelligence (GenAI) is an innovative tool for creating learning tasks that are tailored to the individual needs of students and respect the principles of STEM education. The results of the research show that GenAI tools, such as AI assistants GPTs and Gems, can generate high-quality STEM learning tasks of varying difficulty that are applicable in secondary school teaching. The generated tasks were analysed in terms of integration of STEM disciplines, demands on teaching aids, time consumption, teaching strategies, and development of students’ skills for life in the twenty-first century. These tasks have been found to not only develop students’ cognitive skills but also support their abilities needed for twenty-first century life, such as teamwork, collaboration, systems thinking, and creativity.