Skip to Main Content

Generative AI and Copilot: Prompt Engineering

Prompt Engineering

What is prompt engineering?

Prompt engineering refers to the different ways you can instruct a gen AI tool. A clear and precise prompt is more likely to provide a relevant output. Prompting the gen AI tool to respond with a certain perspective, can help put things into your context and level of learning.

Things to consider with prompt engineering for your studies: 

  • It is important that you only use gen AI as an educational tool, and not as a replacement for learning
  • If you use gen AI for assignments, always apply caution and critically evaluate the output information
  • Remember, gen AI is a text-generation tool, NOT a source of credible information, and can hallucinate and fabricate information
    • Check the output against credible sources for accuracy (e.g. Library Search or academic databases)
    • Verify references (if provided), and all information by identifying credible sources to back up the generated output
    • Undertake lateral research to verify and build on the information provided in the output to learn more
  • Be aware of 'confirmation bias' in the output and inaccurate responses

 

Prompting using 'RICE'

Applying a structure to your prompt can produce a more relevant output.

Using RICE as a guide, provide as much context, detail and boundaries as possible:

R - Role

  • The level and role the gen AI should assume, e.g. 'a university tutor'

I - Instructions

  • The focus, format, style, intended audience, e.g. 'research terms for a 2nd year Arts university student'

C - Context

  • Details about the task you want addressed; be clear, e.g. 'Climate Change,' and include examples of your own work to help the gen AI tool ('few-shot' prompting)
  • An example of 'few-shot' prompting would be to include your own brainstormed search keywords and prompt the tool to add any you may have missed

E - Expectations

  • State specific output requirements, e.g. 'help me identify research gaps,' 'do not hallucinate if information is unknown,' 'be accurate and provide links to sources for verification.'
  • It is important not to request for the gen AI tool to output full assessment material as this would be Academic Misconduct. It is only an educational tool, not a replacement for learning.

 

Example 1:

Prompt:

R - Role - Act as a university tutor

I - Instructions -  Help a 3rd year uni student develop a search strategy for a presentation

C - Context - On abstract impressionism and cultural appropriation; include your current search plan

E - Expectation - Give me five points to consider to help me improve this strategy.

“Act as a university tutor for 3rd year visual arts student. I need to develop a search strategy for a presentation on abstract impressionism and cultural appropriation. My current strategy is [include details]. Give me five points to consider to help me improve this strategy.”

 

Example 2:

Prompt:

R - Role – Act as a university tutor in nursing and health sciences

I - Instructions – Help a 2nd or 3rd year nursing student plan and structure a 2,000-word evidence-based case study

C - Context – The case study focuses on a patient with type 2 diabetes and includes nursing interventions, patient education, and evidence-based practice

E - Expectation – Provide five key planning tips to help structure and develop a strong, well-supported case study

Full Prompt:

“Act as a university tutor in nursing and health sciences. I’m a 3rd year nursing student preparing a 2,000-word evidence-based case study on a patient with type 2 diabetes. The case study needs to include nursing interventions, patient education, and evidence-based practice. Can you give me five key planning tips to help me structure and develop a strong, well-supported case study?”

Watch: Prompt engineering.

For more detail on prompt engineering see the Academic Skills @ ECU page on Understanding Generative AI.

Prompt Engineering for Gen AI Image Generators

Prompt engineering for GenAI image generators is similar to engineering for text-based GenAI, except the Context is less important, and the Task and Outcome parts are more visual. Consider:

  • Composition and Framing: “Close-up,” “Low angle,” etc.
  • Style and Technique: “Art Deco,” “1920s,” etc. Or “Black and white,” “Photography,” etc.
  • Context and Vibe: “Studio shoot,” “Vibrant,” etc.

All this is in addition to the Content.

For example, the prompt:

Can you make me a close-up photo of a clown in a space suit with dramatic lighting

Will be very different to:

Please make a wide-angle image of a clown in a space suit in Cubist style.

The Academic Skills @ ECU Canvas site has further strategies and examples of how to best prompt gen AI, including Brainstorming prompts, Editing prompts, and how to prompt for useful Feedback.