Abstract:

Moral education is a process that aims to enable individuals to make conscious and consistent moral decisions by understanding fundamental ethical distinctions, such as right–wrong and good–bad (Nucci, 2001). Values education, on the other hand, refers to a broader framework that aims to help individuals internalise social, cultural, and personal values ,thereby, developing into responsible, respectful, and harmonious members of society (Lickona, 1991; Halstead & Taylor, 2000). When considered together, these two areas form a critical foundation for students to develop both their ethical reasoning skills and to exhibit behaviours consistent with the values of the society in which they live.

However, research shows that young people in secondary school do not always feel comfortable expressing the questions they have in these areas. For example, studies examining young people's behaviour in relation to social norms reveal that students act out of concern for ‘what their peers will think’ when expressing their thoughts on ethically or socially sensitive issues ;therefore, they suppress their real questions (Ommundsen et al., 2023). Educators who examine moral inquiry in the classroom emphasise that students' reluctance to ask existential or moral questions stems from these questions being perceived as ‘risky,’ ‘personal,’ or ‘inappropriate’ in the classroom, and that teacher-student power dynamics are decisive in this process (Fenstermacher, 1992). These findings indicate that it is quite common for secondary school-aged young people to avoid asking direct questions about moral or ethical issues for social and emotional reasons.

Therefore, the existence of a safe, impartial and accessible system where students can freely ask questions without fear of judgement, exclusion, or misunderstanding is of great importance in terms of both values and ethics education. Such a system will deepen and support the educational process by enabling students to express their suppressed curiosity, develop healthy insights on ethical issues, and actively participate in values education.

In this study, an artificial intelligence-supported system that utilises reliable, non-judgmental resources is constructed in values/moral education. The system employing Retrieval-Augmented Generation (RAG) model is designed to answer student questions on ethics and moral reasoning using reliable, non-judgmental guidance. The system processes the 9th grade textbook "Adabımuaşeret" (Bala et al, 2024) prepared by the Turkish Ministry of National Education as its core resource, addressing the need for an unbiased assistant. The queries are routed through a retriever pipeline and processed by the GPT-4o-mini model, with the chain logic implemented using PromptTemplate and RunnablePassthrough within LangChain. To evaluate performance, a set of ethical questions (inputs) are collected from secondary school students, and the correct answers are determined by experts. In addition, a set of out-of-context questions is built to measure the performance of the system. The performance is measured, using well-established natural language processing metrics: accuracy, defined as the proportion of correct responses to total questions, and relevance, which measures how closely the generated answers align with the user’s query. These metrics guided the assessment of system quality and ensured that the model not only produced correct outputs but also maintained strong contextual alignment.

Keywords:

Values education, artificial intelligence, RAG model, morality, ethics.