Generative AI tools can be used to assist instructors with creating and adapting educational materials such as quiz questions, problem sets, case studies, examples, personalized learning experiences, diagrams, audio files, glossaries, and more. The prompts below serve as a template for creating a variety of activities.

Strategies for Responsible Use

Generative AI should not be seen as a replacement for educators, but rather a collaborator that can assist with the co-creation of educational materials. There are a number of limitations and ethical concerns surrounding the use of Generative AI. You are ultimately responsible for the output you use. Below are some strategies you can follow for responsible use:

  1. Accuracy: Generative AI tools may produce content that is misleading or incorrect. Ensure you have the expertise to verify that the output is accurate, appropriate and useful.
  2. Bias: AI models are trained on extensive data sets from a variety of online sources, and thus reflect the same biases that we see in society. Review the output for bias, including content or images that reinforce cultural or societal stereotypes.
  3. Copyright: There is a lack of transparency surrounding the source of data used to train AI models, but we know many have been trained on copyrighted materials without the permission of copyright holders. Include proper attribution when possible and avoid inputting copyrighted materials. Consider being transparent about your use of Generative AI by including a declaration that includes what content was generated, what tools was used, and how you used it.
  4. Privacy: Data input into Generative AI tools is often stored outside of Canada and used to further train the model. Avoid entering personal or confidential information, especially those of your students. Look for tools that can be run locally on your computer, store data within TRU (such as Copilot), and have the option to disable the use of your data for training.
  • Assign Persona
    You are an expert/instructor in _ and Learning Design
  • Construct Output
    Specify the number of questions, format
  • Tailor the Details
    Indicate the level, context, distractors, learning outcomes, taxonomy etc.
  • Offer Examples
    Provide sample question and answers to guide the AI
  • Refine
    Evaluate the output yourself, but you can also ask AI to refine. For example, “Act as an expert in _ evaluation and critique these questions. What criteria did you use? Rewrite based on this critique.”

Adapted from Lucas Wright, under a Creative Commons Attribution-NonCommercial 4.0 International Public License

Prompt:

You are an instructor and expert in [subject area]. Please create [#] [question type, such as multiple choice or true/false] questions with [3-5] alternatives for each question suitable for [target audience]. Focus specifically on [topic, problem etc.]. Questions should be based on the following learning objectives: [insert learning objectives] Include plausible distractors that reflect common misconceptions. Indicate correct answers with an asterisk. An example question is: [insert example question].

Refinement Prompt:

Act as an expert in [same subject area as above] and critique these questions to ensure they align with the indicated learning outcomes. What criteria did you use? Then: rewrite based on this critique.

The Aiken format lets you create multiple-choice or true-false questions using a simple, human-readable format that you can save as a plain text file and import into a Moodle course. Generative AI can be used to generate questions in the correct format, or you can ask it to reformat existing questions.

Prompt:

Please create a multiple-choice quiz on [topic]. Create [#] questions, each of which has four possible answers. The format of the output must match this example:

What is the correct answer to this question?
A. Is it this one?
B. Maybe this answer?
C. Possibly this one?
D. Must be this one!
ANSWER: D

The question is on the first line. The possible answers follow beneath this, each on their own line. The answers must be preceded by a capital A, B, C, or D and a period. The last line must begin with the word ANSWER in all caps, followed by a colon and the letter matching the correct answer. Please separate each question with an empty line. Please output in a markdown code block. The output should not include any additional formatting, such as question numbers, references or superscript numbering for footnotes.

Next Steps:

Save the output as a .txt file. You can then upload the file to your course question bank. See “Is there a quicker way to add multiple choice questions to the question bank?”

Adapted from AI Prompt Library. Vancouver Community College entre for Teaching, Learning, and Research under CC BY-NC-SA.

Generative AI tools can be used to easily generate or reformat existing questions or content to follow an H5P-friendly format. This means that you can simply copy the output and paste in into the “Textual” field of your H5P activity rather than having to manually fill out the default form fields.

Try It!

The H5P Activity Generator (opens in a new tab) is a custom GPT that is already trained to format its output in an H5P-frinedly format. This means that you only need to focus on your topic prompt, indicate what type of activity you would like to create, and paste the output into your H5P activity. Custom GPTs require a ChatGPT account to access. If you do not have an account, you can adapt the example prompts below to achieve the same output.

Activity TypePrompt
Single Choice Set
Create questions with one correct answer.		Create a multiple-choice quiz about [topic] appropriate for [level of learning]. Create [#] questions with 4 alternatives for each question.

Output the questions with the question on the first line, the correct alternative on the next line and distractors on the following lines. Separate each question with an empty line. Do not add any labels, formatting or numbering. Do not assign letters to the alternatives.

Follow the described format precisely as I will be pasting the output into another tool that expects this format. Output the answer as preformatted code.

See Example
Quiz (Question Set)
Create a sequence of various question types		Create a multiple-choice quiz about [topic] appropriate for [level of learning]. Create [#] questions with [#] alternatives per question.

For each distractor add explanation about why the distractor is not a correct alternative. Output the questions with the question on the first line, the correct alternative on the next line with an asterisk in front of it and distractors on the following lines. Each distractor should be on their own line. Suffix each distractor with ‘::’ and the explanation for why it is not the correct alternative. Separate each question with an empty line.

Do not add any labels, formatting or numbering. Do not assign letters to the alternatives.
Follow the described format precisely as I will be pasting the output into another tool that expects this format. Output the answer as preformatted code.

See Example
Summary
Create tasks with a list of statements		I am using H5P to create interactive summary tasks. H5P presents statements in random order where one statement is correct and the others are false. The user is to pick the correct statement and then a new set of statements is presented for the user to pick from until the user has created a summary.

Please create a summary of [topic] appropriate for [level of learning]. Write it as [#] statements, but for each of the statements also create three incorrect statements so that the reader is challenged to find the [#] correct statements.

Use this format: Write the correct statement on the first line and add three distractor statements on the next three lines. Use an empty line to separate sets of statements. Do not add any labels, formatting or numbering and no extra empty line below the correct statement. Output the answer as preformatted code.

See Example
Drag the Words
Create text-based drag and drop tasks		I am creating a learning activity where students drag keywords from a word bank into spaces in a paragraph of text. Keywords should be important words or concepts.

Create an H5P drag the words task about  [topic] suitable for [level of learning]. Include [#] paragraphs and [#] droppables per paragraph. The output should be at least [#] words. Add explanations. Here is the format you need for follow:

Droppable words are added with an asterisk (*) in front and behind the correct word/phrase.
You may add feedback for each droppable to be displayed when a task is completed. Use ‘\+’ for correct and ‘\-‘ for incorrect feedback. Feedback should be enclosed between the two asterisks along with the droppable word itself. Droppables can also include a hint. Hints follow directly after the droppable and are preceded by a ‘:’ sign.

Here is an example:
H5P content is *interactive\+Correct! H5P is all about interactive content.\-Incorrect, H5P is actually all about interactive content.* and can be used to share rich content between websites.
Do not add any labels, formatting or numbering. Output the answer as preformatted code.

See Example
Fill in the Blanks
Create a task with missing words in a text		Create a fill in the blanks quiz about [topic] suitable for [level of learning]. The quiz should have [#] paragraphs with [#] blanks per paragraph.

The blanks should be important words or concepts. Each blank should be prefixed and suffixed with asterisks. If the blank can have multiple acceptable synonyms as answers use a forward slash between them and no extra whitespace and add any such acceptable synonyms between the asterisks. Where suitable also add hints to what the correct blank is by prefixing the hint with a ‘:’ and adding the hint after the correct word(s) and before the asterisk that suffixes the blank.
Output the answer as preformatted code.

Here is an example: “Office workers typically use a *desk/table:Furniture that has a flat top to place things on* to put their computer on”

See Example
Adapted from ChatGPT Prompt Recipes under Creative Commons Attribution 4.0 International