This thesis explores the advantages and risks of introducing an AI assistant for students in higher education, focusing on a database design course at University West. By developing and evaluating AI assistants trained on historical course material such as lecture notes, assignments, and syllabi, the study addresses limitations in traditional academic support systems, as well as the potential benefits and risks of AI assistants. Utilizing a qualitative methodology, the research involved semi-structured interviews with five students and three teachers to investigate four key areas: how AI assistants could potentially enhance access to course material, the amount of training data required for a competent AI assistant, the reliability of AI-generated responses and student perceptions compared to human instructors. Results reveal that AI assistants improve accessibility through constant availability and efficient resource aggregation, though risks of superficial learning and over-reliance are a concern. Competence hinges on balanced, high-quality training data, while reliability requires robust validation to mitigate inaccuracies like linguistic errors or hallucinations. Compared to human teachers, AI offers efficiency and focus but lacks emotional depth and contextual nuance. This study contributes to educational technology by highlighting AI’s potential as a complementary tool in higher education, while emphasizing the need for careful integration to optimize its benefits and address its risks and limitations.