The integration of artificial intelligence (AI) in education is rapidly transforming teaching and learning processes, with large language models (LLMs) such as ChatGPT, Microsoft Copilot, and Google Gemini leading the charge. This study investigates the impact of LLMs on core academic subjects in Ghanaian tertiary institutions, focusing on their effects on teaching effectiveness, student engagement, and learning outcomes. Using a qualitative approach, we gathered insights from both educators and students in disciplines such as computer and information security, programming, business intelligence, and research methods across institutions like the Ghana Institute of Management and Public Administration (GIMPA), University of Ghana, and University for Development Studies. The findings reveal that LLMs significantly enhance teaching by streamlining content preparation, facilitating interactive learning, and offering personalized, real-time feedback. Students reported increased engagement and improved comprehension, which contributed to better academic performance. However, challenges such as the financial burden of accessing premium versions, technical barriers like unreliable Internet access, and risk of student overreliance on AI tools were identified. Additionally, the study highlights ethical concerns surrounding data privacy, algorithmic bias, and equitable access, particularly in resource-constrained environments. To address these issues, the need for localized AI solutions tailored to specific cultural and linguistic contexts is emphasized. Furthermore, the research proposes the development of robust metrics to measure system-wide participation and assess the effectiveness of AI-driven educational initiatives. These metrics are essential for informing future AI integration strategies in educational settings. The study concludes by calling for further research to refine these findings through quantitative analysis, aiming to provide deeper insights into the optimal use of LLMs in diverse educational contexts.