Writing AI Prompts in Post-Secondary Classrooms
Actionable ideas for students and faculty
Table of Contents
Estimated read time: 6 minutes
What?
Artificial intelligence (AI) tools, such as ChatGPT, use a prompting interface to generate outputs. Essentially the user asks the system a question and the system produces an answer based on what is asked and it's training.
Getting good, usable results from AI tools often relies heavily on inputting solid, well-articulated prompts. In the last three years, more and more resources have been developed around AI prompt writing and how to teach prompt writing to post-secondary students.
Writing prompts for text output can vary from writing prompts for image output. The information below focuses on text output (e.g. ChatGPT).
So What?
General Prompt Writing
Before we can teach our students how to write a good prompt, we need to experiment ourselves with writing prompts. According to Seneca Polytechnic, there five main components to good prompting:
- Context: provide relevant context including the roles of both AI and the user in the task, outlining your requirements, and specifying the anticipated results.
- Instructions: include explicit instructions for the desired behaviour and format (e.g. step-by-step, multiple-choice questions with answers, ..)
- Examples: provide example inputs so the AI can understand the desired pattern or formats for its response (e.g. relevant keywords or descriptors, reference text, abstract of an article, etc.)
- Outcome Indicators: direct the AI to generate a response for a given style, tone, format and/or scenario.
- Test and Tune: take time to test and refine your prompts until you get to your desired response. Prompt techniques will change depending on task, AI tool, etc.
To read more and see some common prompt examples and tips, visit Using GenAI Tools - Prompting - Generative Artificial Intelligence by LibGuides at Seneca Libraries (senecapolytechnic.ca) [new tab].
Your turn: use an AI tool (e.g. ChatGPT or Co-pilot) and create some prompts to get a feel for how the system responds.
This material is adapted from Using GenAI Tools - Prompting - Generative Artificial Intelligence” by LibGuides at Seneca Libraries (senecapolytechnic.ca). The website is licensed under the CC BY-NC-SA 4.0 license.
Research Prompt Writing
Sheridan College's Library and Learning Services [new tab] provides tips and examples for writing research prompts. Using the steps above, here are some example prompts: (Sheridan College, 2024)
- Brainstorm prompt: "I'm writing a research paper for [course] and I need help coming up with a topic. I'm interested in topics related to [subject]. Provide a list of 10 topic ideas related to that."
- Narrowing down of topics prompt: "I like the topic of [topic] in [subject]. Suggest some sub-topics to explore further."
- Writing research questions prompt: "Provide some research questions for the [sub-topic]." Also try: "Act as a research expert. Provide some feedback on the following research question: How did the [idea] of [subtopic] reflect [idea] of the [subject/course]? Provide some suggestions for improving the research question."
- Identify key words prompt: “Using the research question [insert], provide some keywords to use when searching library databases.” Also try: "Construct a few Boolean search strings I can use when researching [topic] in library databases."
Your turn: use an AI tool (e.g. ChatGPT or Co-pilot) and develop a research topic for a presentation/report/infographic. See how far you can get with your research topic development.
Now What?
Once we get an AI tool to provide a response, the response needs to be evaluated. Have you tried the SIFT method? It's all about recontextualizing the material.
- STOP - Do you know the website or source of the information? What is the reputation of the claim and the course?
- INVESTIGATE the source - Know what you are reading before you read it.
- FIND better coverage - Look for better or more trusted sources or additional sources on the claim in question. Is there a consensus?
- TRACE claims, quotes, and media back to the original context - Figure out the before and after context to get a bigger picture of the claim.
Your turn: use the AI output from your research inquiry (above) and evaluate (SIFT) what information or suggestions have been provided.
This material is adapted from Evaluating GenAI Content - Generative Artificial Intelligence - LibGuides at Seneca Libraries (senecapolytechnic.ca). The website is licensed under the CC BY-NC-SA 4.0 license.
Want to read more?
- Navigating the biases of AI is described in this Faculty Focus article (free sign up may be required): Mitigating Hallucinations in LLMs for Community College Classrooms: Strategies to Ensure Reliable and Trustworthy AI-Powered Learning Tools - Faculty Focus | Higher Ed Teaching & Learning
- A Teacher's Prompt Guide to ChatGPT aligned with 'What Works Best' (wordpress.com) [PDF] examines prompt ideas for setting high expectations, explicit teaching, effective feedback, use of data to inform practice, assessment, classroom management and collaboration. VERY HELPFUL when using ChatGPT to support lesson development.
References & Attributions
Sheridan College - Library and Learning Services. (2024, June 30). All guides: Using generative artificial intelligence for research: Creating text prompts. Creating Text Prompts - Using Generative Artificial Intelligence for Research - All Guides at Sheridan Library & Learning Services.
Where noted, the material in this article is adapted from:
- Using GenAI Tools - Prompting - Generative Artificial Intelligence” by LibGuides at Seneca Libraries (senecapolytechnic.ca). The website is licensed under the CC BY-NC-SA 4.0 license.
- Using GenAI Tools - Prompting - Generative Artificial Intelligence” by LibGuides at Seneca Libraries (senecapolytechnic.ca). The website is licensed under the CC BY-NC-SA 4.0 license.