Generative AI in Higher Ed

A prompt is the question or set of instructions the user provides to the large language model (LLM) to have it perform a task. The LLM then returns its results to the user in a natural language response. The basic components of an engineered prompt will identify what you want the LLM to do, context for why it is needed, and your expectations for output formatting and tone. The more specific and curated a prompt is, the more tailored the results will be for the user.  

Prompt Engineering Basics

• Use plain language and be specific about what you want the LLM tool to do.
• Provide some background or context for the LLM so that it can produce results specific to your circumstances. 
• Specify your intended audience and/or the tone or style in which you want the results to be generated (e.g., professional or formal language; friendly and casual language; in language that will be easily understood by beginning English language learners).
• Refine results by asking for variations or specific adjustments to results.  

The CLEAR Framework can be used in order to develop effective prompts: 

• Concise: Use brief, clear, specific language.
• Logical: Use structured, coherent, ordered language.  
• Explicit: Include specific guidelines for the process and output you are requesting.
• Adaptive: Observe your results closely and respond flexibly. Experiment with your prompts.
• Reflective: Are your results relevant, accurate, and complete? Continue to evaluate and revise your prompts to further develop your prompt engineering skills.

You can read more about the CLEAR framework here: Lo, Leo S. (2023). "The CLEAR Path: A Framework for Enhancing Information Literacy through Prompt Engineering." The Journal of Academic Librarianship, 49(4). https://doi.org/10.1016/j.acalib.2023.102720.