Development options for artificial intelligence supported intelligent tutoring systems via the integration of the ChatGPT

Abstract

In the era of the second revolution of intelligent tutoring systems, the integration of artificial intelligence can bring about a learning process adjusted to the personal needs and preferences of students, including language teaching, sinse AI-based systems can assist in identifying and addressing individual language skills and deficiencies. While these systems can be divided into four inherently AI-based components (learner model, domain model, pedagogical model and user interface) facilitating the embedding of cognitive functions, the options hidden in the integration of big language models have not yet been fully utilized. The first step in the integration of ChatGPT developed by OpenAI is the thorough identification of the goals of the education process followed by the allocation of the appropriate API (Application Programming Interface) service to the given process (question and answer, content summary, sentiment analysis). At the same time it is worthwhile to improve the ITS architecture to maximise the ChatGPT provided options. One result of our research and development effort is a further improved architecture integrating an additional component, the central organizer. This unit is responsible for the matching or allocation of the ChatGPT’s API parameters to the given educational content and objective. The integration of ChatGPT into intelligent tutoring systems poses numerous challenges such as the protection of personal information, the related technological problems along with the clarification of ethical issues (Ara-tó–Balázs 2023). Addressing such concerns is necessary for the realization of our original objective guaranteeing that the respective technological achievement would effectively support the implementation of said goals.