Prompt Design For Beginners

Welcome to the Prompt Design for Beginners workshop! 

No prior knowledge is required, just bring your curiosity. Expect a friendly journey into prompt design with exciting examples featuring astronauts, space exploration, song titles, zombies, and dinosaurs. All of the important things. 

This workshop will help you get started on your AI (artificial intelligence) literacy journey. 

• AI literacy extends beyond mastering the skill of crafting prompts for tools like ChatGPT-3.5. It encompasses the ability to critically assess their outputs, delve into the intricacies of their mechanisms, and reflect on their ethical and societal implications (Lo, 2024)

This workshop was created by Laura Briggs during her term as the UVic Data Science Librarian. Questions about this workshop can be addressed to askus@uvic.ca. 

Reference

Lo, L. (2024). Evaluating AI literacy in academic libraries: A survey study with a focus on U.S. employees. University of New Mexico Digital Repository. https://digitalrepository.unm.edu/ulls_fsp/203/

If you only have five minutes to invest in learning how to write effective prompts then I suggest choosing one of the following two learning strategies.

• Read the Use Delimiters module to learn how to structure a prompt. This is the easiest technique to improve the model's responses.
• The Putting It All Together module contains a detailed summary table outlining all of the prompt writing strategies and tips covered in the workshop.

Feel free to revisit this workshop and explore different modules whenever you have some spare time. You don't have to follow a strict linear pathway through the workshop's content. Instead, select a strategy that interests you and integrate it into your prompt design process.

The majority of the modules are organized into the following sections:

1. A summary of the key points of the module.
    
2. A diagram that provides an overview of the examples used in the module.
    
3. Each module includes two or more examples of prompts.
   The examples are structured:

• Exposition from Laura explaining the main concepts.
• Details highlighting important aspects about the upcoming prompt. 
• The prompt.
• ChatGPT's response.

4. Modules conclude with a small activity, Your Turn!, so that you (ie. the learner) can check your understanding of the module's key points. There are a variety of question formats to keep everyone engaged such as:

• True or False
• Arrange in order
• Multiple choice
• Open ended prompt writing suggestions with Laura's solution to compare with. 

In an example, an interaction between the prompt writer (ie. Laura) and ChatGPT (ie. the model) is structured as shown below. ChatGPT's response is displayed in blue text to help distinguish its words from Laura's prompt.

• Laura:
  The prompt.
  (in black text)
   
• ChatGPT:
  A response.
  (in blue text)

Laura's prompt and ChatGPT's response are directly copied into the module, making it simple for learners to run the same prompt in ChatGPT if they choose. Note that the response you receive from ChatGPT will not be exactly the same as the answer Laura received. ChatGPT is tricky that way!

• Here is an example of an interaction between Laura and ChatGPT, as it would appear in a module:

Prompt In ChatGPT-3.5: 

Laura:

Evaluate My Text for clarity, accuracy, and ease of understanding.

My Text:
A prompt is the input you provide to ChatGPT, whether it's a question or a command. It instructs the model on what action to take or what information to provide during your conversation.

ChatGPT:

Your text is clear, accurate, and easy to understand. It effectively explains what a prompt is in the context of interacting with ChatGPT.
Well done! If you have any more text you'd like me to evaluate or if you have additional questions, feel free to let me know, Laura.

• The screenshot below shows how the interaction actually appears in ChatGPT-3.5.

Image Caption: A screenshot showing what an interaction between Laura, the prompt writer, and the model, ChatGPT, looks like in ChatGPT-3.5.