The rapid diffusion of artificial intelligence (AI) into academic library services has widened the gap between libraries in high resource environments and those in the Global South. Nigerian public university libraries, serving an expanding higher education population, face a compounded set of barriers described in this paper as a triple threat: unreliable internet connectivity, chronic power instability, and costly foreign currency subscription models. This study adopted a mixed methods design involving ninety respondents drawn from ten public universities in North Central Nigeria, comprising ICT directors, university librarians, systems librarians, and digital resources officers. Data were analysed using descriptive statistics, a Weighted Average Index, and inferential tests. Findings show that bandwidth availability strongly predicts AI adoption readiness and that federal universities significantly outperform state universities on composite readiness scores. The offline AI powered reference chatbot emerged as the most desired application. The paper proposes a Frugal AI Framework built on offline language models, open-source systems, solar power, and localised African training data, and recommends targeted policy action by professional and funding bodies.
Keywords: Artificial Intelligence, ICT infrastructure, Library, Nigerian university libraries, digital inequality, frugal innovation, offline AI
