Gen AI breaks down words into tokens that it can process. 750 words are approximately 1,000 tokens. AI tools have limits on how many tokens they can process. ChatGPT is limited to 4096 tokens. This token limit applies to both the input (your prompt) and the output (ChatGPT's response).
The CLEAR Framework for Prompt Engineering was developed by Leo S. Lo, Dean and Professor of the College of University Libraries and Learning Services at the University of New Mexico. |
Prompting is used to instruct an AI to do a task. The better prompts (aka questions and inputs) you use with generative AI, like ChatGPT, the more relevant and meaningful responses (outputs) you'll receive!
Concise - Use concise and clear prompts.
X mindfulness
X Tell me about mindfulness.
List three examples of mindfulness strategies for college students [instruction prompt].
Logical - Maintain a logical flow and order of ideas.
List three examples of mindfulness strategies for college students. Follow-up:
Write a brief summary about how mindful eating helps with college stress. Follow-up:
1. Savoring a bite [few-shot/examples prompt]
2. Chewing slowly
What other examples are there?
Explicit - Give precise instructions on what the response should include and how it should be expressed.
Imagine that you just got married [role prompt]. You want to practice mindful eating with your husband at dinner time. What's the best way to approach this topic with him?
NOTE: If you need to break up your prompts into paragraphs, start a new paragraph with the combined keystroke [shift] [return/enter]
Adaptive - Be flexible! Experiment with different prompts and adapt to the AI responses.
Reflective - Reflect on the AI-generated content and evaluate the responses. Confirm the accuracy, relevance, and completeness of the responses.
One more consideration! To have a conversation with generative AI tools, ask AI to ask you a question related to the topic. Answer the question to keep the conversation going.