Abstract:

Explainability in Artificial Intelligence (AI) has become essential for promoting transparency and trust in AI-based systems. Despite its growing relevance, the integration of explainability concepts into Requirements Engineering (RE) education remains limited, hindering the preparation of professionals capable of eliciting explainability requirements in user-centered systems.

This paper presents an educational study that investigates the impact of using the End-User-Centered Explainable AI (EUCA) framework in RE education, focusing on the elicitation of explainability requirements in AI-based systems. The proposed approach combines the introduction of explainability concepts with the practical application of the framework, enabling students to understand and apply these concepts from an end-user perspective.

The study was conducted with undergraduate and graduate students enrolled in RE courses across multiple classes. The teaching process was structured around the practical use of the framework, including the analysis of AI system scenarios and the construction of prototypes aimed at eliciting explainability requirements.

To evaluate learning, a pre- and post-test composed of Likert-scale and open-ended questions was applied to assess both conceptual understanding and students’ ability to elicit explainability requirements. The analysis considered the comparison between results obtained before and after the intervention, as well as the qualitative interpretation of participants’ responses.

The results, based on the comparison between pre- and post-test, allow for analyzing the impact of using EUCA on students’ ability to elicit explainability requirements in AI-based systems from an end-user perspective. Qualitative findings complement these results by showing how students structure their reasoning and relate explanation needs to appropriate forms of explanation.

As a contribution, this work presents and evaluates a practical approach based on the EUCA framework for teaching the elicitation of explainability requirements in AI-based systems, providing empirical evidence of its impact on students’ learning. The findings highlight the role of user-centered prototyping as an effective pedagogical strategy and discuss implications for Software Engineering education in AI-related contexts.

Keywords:

Requirements Engineering Education, Explainability in AI, Software Engineering Education.