An Exploratory Study of Large Language Model-Based Writing Support for Postgraduate Engineering Students at a South African University

Postgraduate engineering students in South Africa face ongoing challenges in developing comprehensive academic writing skills, which can lead to complications in their research timelines and contribute to attrition rates at tertiary insitutions. Conventional academic writing support programmes typically focus on general academic literacy, often failing to address the specific needs of engineering students. Additionally, limited opportunities to engage with supervisors for comprehensive feedback on thesis writing further exacerbate these challenges. Consequently, there is a critical need to develop tailored academic writing support structures that actively build students' writing abilities, offering high availability and versatility to adapt to individual student needs and a broad range of research topics. This paper reports on an exploratory pilot project that integrated a Large Language Model (LLM) as an AI assistant (AIA), using the Mindjoy online educational platform, into the thesis-writing process for postgraduate students at North-West University's Faculty of Engineering in South Africa. Importantly, the model was designed to prohibit content creation, focusing instead on developing students' core writing competencies. Preliminary findings indicate that the AIA can significantly enrich students' writing processes, potentially leading to improved academic performance and the acquisition of essential writing and communication skills. However, ensuring ethical use while preserving students' ownership of their work requires a careful approach, along with comprehensive user orientation and training.