In Japan, teachers from elementary to high school provide classes according to the Courses of Study, which is revised every 10 years by the Ministry of Education, Culture, Sports, Science and Technology (MEXT). A teaching plan is a document that specifies the contents and learning activities of a class in concrete terms. Each teacher is expected to use a teaching plan that describes educational class design and its methods in everyday classes. However, due to the teachers’ overwhelming daily workload, teaching plans are used only for occasional demonstration classes or class research discussions. In short, they are rarely utilized for ordinary classes. In addition, there are two types of teaching plans: detailed plans and simple plans. Generally, a detailed plan is called a teaching plan, and a simple plan is a brief version of a detailed plan. The simple plan is easier in format than the detailed plan and focuses on the contents and methods of the class and the activities of the students. If teachers can create a simple plan for each class, it would improve the quality of their teaching activities and enable them to manage their classes on a continuous basis. This would function as a quite important educational database. However, in reality, it is difficult for teachers to make a simple plan for every class and utilize it under the current circumstances. If teachers in schools can create the simple plans for daily teaching and utilize them for innovative ideas and updating for class management, the quality of education can be improved even more than before, despite the demands of tough work as teachers. In this research, we particularly focused on leveraging a generative AI to create a simple teaching plan. We constructed a method that enables simple and convenient creation, which basically does not require human intervention, by utilizing a generative AI.

抄録全体を表示

This paper proposes TAK10, a design-oriented framework for prompt-level control of artificial personality in large language models (LLMs). Although LLMs are stateless at the parameter level, TAK10 introduces an explicit state-transition mechanism that modulates personality weights across dialogue turns without modifying internal model parameters. Personality is represented as a constrained weight vector over multiple modes and updated through bounded adjustment, normalization, and periodic resynchronization, enabling consistency with controlled variability. A central component is the E_{score} mechanism, a scoring-based output control model that evaluates responses across dimensions such as reference reliability, consistency, factual stability, validation breadth, and contextual alignment. Through threshold-based regulation, E_{score} supports structured response moderation and enhances transparency. Rather than serving as a statistical inference model, E_{score} functions as a design-level scoring framework that promotes reliability awareness in generative systems. From a data science perspective, the personality weight vector can be interpreted as a constrained state variable on a normalized simplex, and the update rule resembles a bounded discrete-time state-transition system. TAK10 is presented as an architectural contribution, with empirical validation and comparative evaluation reserved for future work.

抄録全体を表示