Introduction to Generative AI: Writing prompts

What is a prompt?

AI prompting is the way that users interact with Generative AI tools.  An AI prompt can be any form of text such as a question, information, or coding, that requires a response.

Gen AI tools are always willing to please and will always provide an answer for you... although it may not be exactly what you are looking for!  To achieve the best results try using one of the following methods.

 

CLEAR framework

Devised by Leo Lo at the University of Mexico, the CLEAR framework is designed to optimize interactions with language models such as ChatGPT.

The CLEAR Framework's five components are:

1. Concise: brevity and clarity in prompts

2. Logical: structured and coherent prompts

3. Explicit: clear output specifications

4. Adaptive: flexibility and customization in prompts

5. Reflective: continuous evaluation and improvement of prompts

For more information about the framework, and examples of prompts:

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.

 

How would ChatGPT like us to use prompts?

Just for fun we asked ChatGPT how effective prompts should be crafted.  Here is the response:

Crafting effective prompts for accurate Generative AI responses requires clarity, specificity, and structure. Here are some key instructions to follow:

1. Be Clear and Specific

  • Clearly define what you want. Vague prompts lead to vague answers
  • Example: Instead of "Tell me about AI," ask "What are the key differences between machine learning and deep learning?"

2. Provide Context

  • Include relevant background information to guide the AI.
  • Example: Instead of "Explain photosynthesis," ask "Explain photosynthesis in simple terms for a high school biology student."

3. Define the Format of the Response

  • Specify the structure you prefer (e.g., list, table, summary, explanation).
  • Example: "Summarize the history of quantum computing in three bullet points."

4. Set a Desired Length or Depth

  • Indicate how detailed the response should be (brief, in-depth, technical, layman-friendly).
  • Example: "Provide a one-paragraph summary of blockchain technology."

5. Use Examples When Necessary

  • If you have a specific style or approach in mind, provide an example.
  • Example: "Write a professional email requesting a meeting, similar to this: [insert example]."

6. Specify Constraints or Exclusions

  • If you need a response within limits, mention it (e.g., word count, specific focus, exclusions).
  • Example: "Explain climate change without using scientific jargon."

7. Request Step-by-Step Explanations for Complex Topics

  • If a topic is complex, ask for a breakdown.
  • Example: "Explain how neural networks work in simple steps."

8. Use Role-Based Prompts for More Contextual Answers

  • Ask the AI to respond from a specific perspective.
  • Example: "As a university professor, explain academic integrity in AI-assisted writing."

9. Iterate and Refine

  • If the response isn’t accurate, refine the prompt by adding details or specifying corrections.
  • Example: If "Explain AI ethics" is too broad, refine it to "What are the ethical concerns of AI in hiring decisions?"

(OpenAI, 2024)

OpenAI. (2024). ChatGPT (Feb 4 version) [Large language model]. https://openai.com/chatgpt