Skip to Main Content

Artificial Intelligence (AI)

Crafting Effective Prompts

prompt is typically a question, request, or topic posed by a human researcher to an AI tool. 

  • Good prompts help the AI help you. A well-constructed prompt guides the AI towards relevant and useful information. Poorly worded prompts can generate misleading or unhelpful results.
  • Good prompts are cheap(er). Many AI tools charge per query. A well-crafted prompt can reduce the number of searches needed, saving you both time and money.

The CRISPE Framework

CRISPE

CR: Capacity and Role. What role do you want the AI to play?
I: Insight, Provide necessary information or context. 
S: Statement. What do you want the AI to do?
P: Personality. Define the AI's tone (style or manner in which you want the AI to answer you).
E: Experiment. Ask the AI to provide multiple answers for you.

Example Prompt: 
I have received an email requesting a job interview for a gym assistant position at a college (I). Playing the role of a college student with no prior job experience (CR), draft a response saying I am very interested in the position and am available for an interview (S). The tone should be excited but professional (P). Draft 3 example responses (E)


Adapted from  "Unleashing ChatGPT's Power: A Case Study on Optimizing Information Retrieval in Flipped Classrooms via Prompt Engineering," in IEEE Transactions on Learning Technologies, vol. 17, pp. 629-641, 2024, doi: 10.1109/TLT.2023.3324714.

The CLEAR Framework

Well-crafted prompts are CLEAR - a framework developed by Leo Lo, librarian and professor at the University of New Mexico.

Concise (also Clear) - Focus on core keywords. Keep your prompts concise by eliminating unnecessary words.
  • Example Prompt:  I am trying to figure out if I should be applying to UT and if I would like it there.
    • Issues: The AI doesn't know you or your preferences, there are multiple "UT" institutions, & are you applying to high school? Undergrad? Grad school?
  • Improved Prompt: Identify the top reasons undergraduate students attend UT Tyler.
Logical - Most AI tools analyze relationships between words and concepts so make sure your query is clear and logical. If your question confuses you, it will likely confuse the AI.
  • Example Prompt: Can we make a flu vaccine organically?
    • Issues:  Are you asking about a vaccine made with eggs obtained through organic farming? A vaccine developed through "organic" research? Something else?
  • Improved Prompt: Summarize the most promising vaccine candidates that protect against multiple strains of influenza. Highlight any mention of "organic" methods. 
Explicit - Clearly state your desired outcome. This helps the AI provide relevant results.
  • Example Prompt: What's UT Tyler like? 
    • Issues: What is your comparison for UT Tyler? Do you want a short or long answer? Are you referring to student organizations? Degree program rankings?
  • Improved Prompt: Give me a concise summary of the major strengths and weaknesses of UT Tyler.

The last two components explain how to refine your prompt based on the AI's initial response. 

Adaptive Incorporate keywords or topics used in the AI response. Refine your prompt by adding or removing concepts.
  • Prompt 1:  Why doesn’t Tyler have a Metro station?
    • Answer: (includes "geological obstacles")
  • Adapted Prompt:  What are the engineering challenges and geological obstacles to constructing a Metro station in Tyler?
Reflect - and Refine. 
Does the AI's response make sense? Is it based on current research (if applicable)? Is it accurate or does it contain misinformation/hallucinations? Is it complete?  If necessary, create new prompts to address unanswered questions.
  •  Prompt 1: Give me a concise summary of the major strengths and weaknesses of UT Tyler.
  • Refined Prompt: Give me a concise summary of the major strengths and weaknesses UT Tyler, from the perspective of a first-generation college student.

Read more about the CLEAR framework in Lo, L. S. (2023). The CLEAR path: A framework for enhancing information literacy through prompt engineering. The Journal of Academic Librarianship, 49(4), 102720–. https://doi.org/10.1016/j.acalib.2023.102720

Like us on Facebook

Follow us on Instagram

Check us out on YouTube