Evaluating the Effectiveness of AI Tutoring Assistants in Teaching Ethical Principles to Accountancy Students

Introduction: This research explores the impact of AI tutoring assistants on accountancy students' learning outcomes regarding the EP100 ethical framework. The study evaluates AI tools' potential for personalized learning and ethical education and gathers students' feedback on the AI tutor's effectiveness and question quality. The study is grounded in educational theories like cognitivism, inquiry-based learning, and cognitive load theory. It assesses the AI tutoring assistant's effectiveness in enhancing understanding and application of topics. The AI tool promotes active engagement and schema development through practice questions. It supports inquiry-based learning by encouraging students to investigate and think critically, providing explanations and examples for independent inquiry. Additionally, it manages cognitive load by breaking down complex concepts and offering immediate feedback. Methods: The study employs an experimental design involving 64 SIT accountancy students, divided into control and treatment groups. Participants first took a pre-test assessing their initial knowledge of the EP100 ethical principles before being allocated into experimental and control groups. Both groups attended a short lecture on the EP100 framework, followed by identical in-class activities such as solving ethical dilemmas and discussing case studies. The experimental group used the AI tutoring assistant during these activities while the control group relied on study materials. Finally, participants took a post-test and the results were compared to evaluate the AI tutoring assistant's effectiveness. Mixed methods, including a difference-in-differences estimation was used to analyze the quantitative data while a post-study survey gathered qualitative feedback on the AI tool's adequacy and the quality of generated questions. The survey included a 5-point Likert scale to assess the adequacy of the AI tutor's responses and an 8-item rubric to evaluate the quality of the practice questions. Results: RQ1 explored the AI ...