Paul Wadden, Istvan Zoltan Zardai, Sophie Otsuru, and Hiroaki Umehara. 2024. Creating and Crafting CLIL Content Using AI Tools: A New Frontier. PDF
This article offers one answer to a key question posed by the organizers of the Fifth International Symposium on Academic Writing and Critical Thinking: How can teachers and students make optimal use of AI? It recounts the authors’ steps and struggles at a major Japanese university to use Large Language Models—in this case, ChatGPT—to create level-appropriate, content-based reading and listening passages for a semi-intensive English program. The authors explain how they first identified gaps in content materials in their liberal arts CLIL curriculum, and then underwent the process of crafting prompts that would generate multiple versions of the same passage to meet the needs of students at different ability levels (TOEFL ITP scores ranging from 350 to 550). The authors note how suitable CLIL content in fields such as Economics, Sociology, Natural History, Earth Science, and Health and Medicine are seldom available from commercial publishers, and they explore the promise that AI-generated content has for creating optimal learning materials. They also illustrate the limitations and the pitfalls of AI-generated materials: namely, the occasional failure of AI to follow directions, the rare but real tendency for hallucination of answers, the potential for plagiarism, and the need to carefully check AI-generated questions and answers for errors. The authors observe that the challenges that their teaching staff and students face are similar to those at many other universities across Japan, and they conclude that ChatGPT can be extremely useful when focused on appropriate topics using artfully crafted prompts. AI tools have great promise for generating level-appropriate texts in the right style and register that can then be further adapted for optimal reading and listening materials.
Keywords
AI, CLIL, Content-Based Curriculum, Japanese higher education
AWCT 2024