Abstract:

ChatGPT, an artificially intelligent (AI) chatbot has received a lot of attention recently due to the powerful capabilities it presents to “follow an instruction in a prompt and provide a detailed response”, according to open-AI. Based on large language models’ tools, ChatGPT can interact in a conversational way, and these models are here to stay. Many organisations are and bringing to production similar language models trained or very large corpora of publicly available datasets. This provide us with the capacity to generate realistic text which in many cases is indistinguishable from human generated text. Along with these advancements there are associated challenges, one of which will be explored in this paper. Can text generated by ChatGPT, given questions from an engineering coursework, achieve passing grades? The use of which by students could potentially leads to academic integrity misconduct in higher education. This has been done by taking a coursework on a unit in the engineering department at a University in Scotland and passing those questions as prompts to ChatGPT. The answers generated are marked by human accessors on criteria such as clarity, depth, accuracy, and relevance. For the answers generated by ChatGPT the results are expected to highlight the performance on our selected criteria, and their implications in coursework design. The research paper would be concluded with some reflection on the idea of ‘embracing’ ChatGPT and designing coursework or online exams based on the Bloom’s taxonomy.

Keywords:

ChatGPT, academic integrity misconduct, large language models, online assessments.