Key points are not available for this paper at this time.
Abstract This practice paper introduces a framework to enhance the practical skills of undergraduate engineering students in generative AI technologies. Our goal is to transform students from users of generative AI software into professional creators of new AI technologies. We begin by defining guidelines, emphasizing ethical, responsible, and lawful practices. Then we define the core practical competencies and design learning activities. The framework involves collaboration among undergraduate students, postgraduate tutors, instructors, technicians, legal experts, academic partners, and industry professionals. The framework adopts a three-stage progression approach. At the adoption stage, students become familiar with generative AI software, such as composing textual prompts for image generation with the stable diffusion model. This helps students stay updated on the latest tools and developments in generative AI applications. In the development stage, the focus is on technical training for application programming interfaces (APIs), including language completion, text-to-speech conversion, and semantic search. This covers hands-on learning of using open-source large language models such as Llama2, and commercial cloud services such as Microsoft Azure OpenAI and Google GCP Vertex AI, etc. Techniques like Retrieval-Augmented Generation (RAG) and model fine-tuning are part of this training, equipping students with the necessary skills to enter the final application stage. In this stage, students participate in designing and developing generative AI-based solutions to address real-world problems. Our partnerships extend to the law and social science faculties, where we build customized chatbot solutions. The framework was implemented and evaluated at the Tam Wing Fan Innovation Wing (a.k.a. HKU Inno Wing) 1, a facility within the Faculty of Engineering at the University of Hong Kong dedicated to improving students' practical abilities. Students demonstrate increased awareness of ethical, responsible, and lawful practices in generative AI technologies under the careful guidance of instructors. We conducted an analysis of the written reflections from students in the 2023/24 cohort regarding their understanding of the strengths and weaknesses of generative AI technologies. Furthermore, we assessed how students' awareness of generative AI ethics, responsibility, and legal considerations evolved throughout their reflections. By identifying common blind spots, we gained valuable insights to continually enhance guidance for students at various stages of their learning progress.
Chui et al. (Sun,) studied this question.