Skip to Main Content

AI Literacy

This AI Literacy guide walks through tips and best practices of using generative AI as a research tool, explaining important AI literacy topics like prompt engineering and the ethical considerations that govern AI useage.

Prompt Engineering Techniques

Prompt Engineering Techniques 

Zero-Shot Prompting

What is it?: Zero-shot prompting involves submitting a prompt to a model that does not contain examples or demonstrations. 

Example: Explain the historical significance of Prohibition. 

Few-shot Prompting

What is it?: Few-shot prompting is a method that involves providing a model with relevant examples for the purposes of helping it understand a task and the desired format. 

Example Prompt: 

Q: 7+3=

A: {Ten}

Q: 13+3=

A:{Sixteen}

Q: 1+4= 

A: {Five}

Q: 8+3=

A:

Expected Output: 

{Twenty}

Why This Works: By providing the AI tool with examples of a preferred formatting through this question-and-answer (Q&A) structure, you are essentially teaching the AI how to respond. Instead of answering 20, the AI tool should respond {Twenty} because it will recognize a pattern through the multiple examples given. 

Role Prompting (aka Persona Prompting)

What is it?: Role prompting involves assigning a persona to a model, such as "Act as a teacher. Explain..." Role prompting encourages text styling and imitation and can have an impact on the performance of the model. 

Example Prompt: You are an experienced bookseller who specializes in mystery and thriller novels. You have a talent for understanding readers' tastes based on the books they’ve enjoyed in the past. I’m a reader who loved The Silent Patient by Alex Michaelides, The Lost Man by Jane Harper, and The Hunter by Tana French. Based on those three books, recommend a mystery/thriller novel you think I would enjoy next—and briefly explain why you chose it.

Why This Works: 

  • Defines the AI's role (experienced bookseller)

  • Hints at expertise domain (mystery/thriller specialist)

  • Sets up the task clearly (recommend + justify)

Self-Refine Prompting

What is it?: Self-Refine Prompting is an iterative prompting technique where you prompt a model and then ask it to refine and provide feedback on its output. 

Steps: 

1. Generate the output 

2. Use the output as a new prompt and ask the model to provide feedback and revision suggestions. 

3. Refine the output and repeat the process (if necessary). 

 

Chain-of-Thought Prompting 

What is it?: Chain-of-thought (CoT) prompting involves requesting that the AI model follow a step-by-step reasoning process. This can be as simple as asking the model to think through the prompt step by step before answering. You can also outline the specific steps that the model needs to follow during the thinking process. 

Example of a CoT Prompt: 

Q: Recommend a mystery/thriller book for someone who loved The Silent Patient, The Lost Man, and The Hunter.

A: Let’s think step by step.

Learn More: View examples of chain of thought prompting from Anthropic. Though this page is focused on Claude, these same strategies can be applied to other tools. CoT prompting can come in zero-, one-, or few-shot varieties can can include assigning an AI tool a persona. 


References

Bhatt, B. (2024, September 27). Self-refine prompting. Learn Prompting. 

Kuka, V. (2024, September 27). Role prompting. Learning Prompting

Prompt Engineering Guide. (2025, January 7). Chain-of-thought prompting

Prompt Engineering Guide. (2025, January 7). Few-shot prompting

Syed, M., & Gadesha, V. (2025, January 29). What is zero-shot prompting?

Smiley Memorial Library

Central Methodist University
411 Central Methodist Square
Fayette, MO 65248

660-248-6271
library@centralmethodist.edu