Exploring the use of customised Generative AI in first-year Accounting learning: a case study of a private higher education institution (PHEI).

Abstract

This study explores the integration of a customised Generative AI tool in a first-year Accounting module at a Private Higher Education Institution (PHEI) in South Africa, addressing the empirical gap regarding AI's conditional effectiveness in professional education contexts. Using an explanatory sequential mixed-methods design, quantitative survey data from 149 students was followed by qualitative semi-structured interviews and focus groups with 24 students. The central finding is that student responses were predominantly neutral across multiple constructs (40–47%), reflecting accurate learner assessment of context-dependent, rather than universally positive or negative AI effectiveness. AI supported foundational declarative knowledge acquisition and initial concept learning but proved insufficient for procedural competence development and higher-order assessment preparation, revealing a persistent theory-practice gap. A strong positive correlation between Ease of Use and Perceived Usefulness (ρ = 0.692, p < 0.01) identified usability as a pedagogical gatekeeper, with usability barriers preventing engagement regardless of content quality. Theoretically, findings extend the Technology Acceptance Model into an Educational context (ETAM-E), refine Self-Directed Learning theory for professional education settings, and synthesise these extensions into an Integrated Conditional Effectiveness Framework for AI-Enhanced Learning (ICEFL). Together, these contributions advance understanding of the conditions under which AI supports learning in South African higher education. The study advocates for strategic, evidence-based, and pedagogically intentional AI implementation rather than uncritical technology adoption.