This study aims to validly develop instructional design principles and corresponding AI chatbot prompts required for argumentative writing instruction supported by an AI chatbot. While the rapid acceleration of digital transformation and the widesprea...
This study aims to validly develop instructional design principles and corresponding AI chatbot prompts required for argumentative writing instruction supported by an AI chatbot. While the rapid acceleration of digital transformation and the widespread adoption of large language model–based AI technologies promise unprecedented educational innovation, they simultaneously raise concerns regarding the outsourcing of human-led cognitive processes to AI. Such concerns highlight the risk of weakening the analytical thinking skills expected of future learners and underscore the necessity for AI tools to function not as mere answer-providing devices but as pedagogical tools that stimulate cognitive engagement and promote higher-order thinking. Grounded in this problem awareness, the present study constructed instructional design principles and chatbot prompt structures aligned with Merrill’s First Principles of Instruction, guiding learners through key components of the argumentative writing process, including concept definition, factual verification, value conflict analysis, and position justification.
This study employed a hybrid form of Richey and Klein’s (2007) Type I and Type II design and development research (DDR). Preliminary instructional principles and AI chatbot prompts were developed through an analysis of prior studies, followed by two rounds of internal validation and chatbot usability evaluations conducted with 11 experts. The revised instructional principles and prompts were then applied in a social studies class of 44 eleventh-grade students to examine external validity. Data collected during this process included quantitative measures such as rubric-based argumentative writing scores from pre- and post-tests as well as a survey on perceived effectiveness, and qualitative data including interviews with five students and AI chatbot–learner interaction logs. A mixed-methods approach was employed to analyze these data comprehensively.
The findings indicate that the AI chatbot–based instructional design principles contributed to an overall improvement in students’ argumentative writing skills. Notably, low-achieving learners exhibited greater gains in writing scores, suggesting that the design helped reduce performance gaps. Learners’ perceptions of effectiveness were also highly positive, with all survey domains scoring above the mid-to-high 4 range on a 5-point Likert scale. Interview results revealed that students perceived the chatbot interactions as helpful in clarifying their thinking, structuring logical arguments, and reflecting on their writing. Moreover, the chatbot was found to function not merely as an advisor but as a tool that provided sustained cognitive scaffolding throughout the writing process. Analysis of chatbot–learner interaction codes further showed that learner responses demonstrating higher-order cognition increased as the chatbot progressed through its activation, demonstration, application, and integration stages. The final instructional design principles and chatbot prompts were refined based on the identified need to clarify the complementary roles of AI and teacher feedback and to strengthen elements related to the integration principle within Merrill’s framework.
This study offers a concrete blueprint for how AI tools can support and deepen learners’ cognitive engagement in argumentative writing tasks requiring higher-order thinking. Beyond the conventional effectiveness-focused AI education research, it presents an integrated model that links instructional design with AI prompt development, thereby providing a structural foundation for practical classroom implementation. In particular, the templated prompts serve as a practical guide for teachers who experience difficulties in designing AI-supported writing instruction. By employing a mixed-methods analysis within the external validation phase, the study also provides a nuanced understanding of learners’ experiences and perceptions, capturing the authentic learning processes enabled by AI chatbot–supported instruction.