Prompting Skills
Welcome to the research page for Prompting Skills. here you will find papers that discuss the specific difficulties that experts and lay people alike experience in designing successful prompts for LLM’s and Large Language Agents (LLA’s)
The paper argues that metacognition—our ability to monitor and control thoughts—provides a valuable framework for addressing usability challenges in generative AI (GenAI) systems. By imposing significant metacognitive demands on users, these systems require enhanced monitoring and control. The authors suggest integrating metacognitive support strategies and improving explainability and customizability in
The paper examines non-AI experts’ ability to engage in “end-user prompt engineering” with an LLM-based chatbot tool. Participants struggled with opportunistic rather than systematic prompt design, facing challenges similar to end-user programming. Human instructional expectations and overgeneralization were key barriers, underscoring the need for better tools and prompt literacy.