Generative AI

Writing an AI Prompt

When using AI one must use a prompt in order for the AI to have something to respond to. Prompts usually include: keywords, questions, requests, and phrases to guide the AI to generate a response. The better the prompt the better the output. 

This process is also known as prompt engineering. 

There is no one correct way to engineer a prompt but there are several schools of thought on how to best tackle the process to get the most effective results. Below are some frameworks to keep in mind when designing an AI prompt: 

The University of Notre Dame states that you should focus on Persona, Task, Requirements, and Instructions

• Persona: Prompts starting with “act as ... ” or “pretend to be ... ” will provide responses similar to that of the role which you provide. Setting a specific role for a given prompt increases the likelihood of more accurate information, when done appropriately.
• Task: Be clear about what you want an answer to, what you want the AI generator to do, find, analyze, etc.
• Requirements: Provide as much information as possible to reduce assumptions the generator may make.
• Instructions: Inform the AI generator how to complete the task.
  • University of Notre Dame (n.d.). AI Overview and Definitions. Notre Dame Learning. Retrieved April 8, 2025, from https://learning.nd.edu/resource-library/ai-overview-and-definitions/

The CLEAR Framework recommends that prompts be Concise, Logical, Explicit, Adaptive, and Reflective - Full text access to the CLEAR framework can be found at the link above. For a basic overview see below. 

• Concise: brevity and clarity in prompts
• Logical: structured and coherent prompts
• Explicit: clear output specifications
• Adaptive: flexibility and customization in prompts
• Reflective: continuous evaluation and improvement of prompts

According to Microsoft (specifically talking about Copilot but this method can be adapted for other AIs) you want to focus on these four categories:

• goal - What do you want the AI to do?
• context - Why do you need it? Who is it for?
• expectations - How can the AI best fulfill your request?
• source - What information samples do you want the AI to use?

Again - there is no one correct way to engineer an AI prompt and the process will take refinement. If you're not getting the results you need - rework your prompt and keep trying. 

For more resources on prompt engineering see the following links:

For Educators looking to write specific prompts please see the resource links below. 

If you have suggestions for how to make this page better, please contact Elizabeth Jerow, Library Director (jerow@msoe.edu).