A 3-Step Workflow for Creating Multiple-Choice Questions (MCQs) with AI

Co-written by Nuria Lopez & Stefani Konstanta, Learning Consultants at CBS.

The very time-consuming task of creating multiple-choice questions (MCQs) can be greatly facilitated by Generative Artificial Intelligence (GenAI) tools like Copilot. These tools can generate questions much faster than we can and this is certainly an advantage that most of us want to benefit from.

That said, the quality of GenAI-generated MCQs is not always where we would like it to be. Sometimes the questions are too basic, or the content is not quite accurate. This is why we have built on the original set of ready-made prompts and turned them into an AI-powered MCQ creation workflow — a 3-step process that combines evidence-based question design, AI-generated feedback, and an instant Canvas import option. These applications were themselves built with the help of AI tools and are designed to help you get better results while saving time.

Step 1 – Generate MCQs with the Prompt Generator

A well-crafted prompt is the foundation of good AI-generated questions. The Multiple-Choice Prompt Generator lets you either:

• Quick Copy a ready-to-use static prompt with built-in MCQ writing guidelines.
• Customise & Copy a personalised prompt by filling in your course subject, level, focus, and topics for even better results.

You are teaching a university course on [subject]. This is a [level/type, e.g., introductory, postgraduate] course.

Create [number] multiple-choice questions for students in this course. The questions must be about [topics/concepts], focusing specifically on [e.g., differences between X and Y; application of X; causes/consequences of Y; or subtopics].

Each question must have [3] alternatives labeled a–c: one correct answer and two distractors. Mark the correct alternative with an asterisk (*).

Write stems that:
• Address a single learning outcome
• Use clear, concise wording
• Avoid negative phrasing and ambiguous vocabulary

Write alternatives that:
• Are all plausible and similar in length
• Are parallel in grammar/structure
• Avoid repeating phrases from the stem
• Do not use “All of the above” or “None of the above”

Return the questions in this exact plain-text format:
• One blank line between items
• First line: “Q#. [stem]”
• Next lines: lettered options starting with “a) ” then “b) …”, etc.
• Put a single * at the end of the correct option line (e.g., “b) …*”)
• No explanations, feedback, or extra text

Example (3 options):
Q1. [your question here]
a) [option one]
b) [option two]*
c) [option three]
      

Start typing — your personalised prompt appears here.

Step 2 – Import your quiz to Canvas in a few minutes

With your MCQs finalised, paste them into the Canvas QTI Converter. The tool automatically formats them into a QTI .zip file ready for import into Canvas — no manual retyping required.

Q1. Which metric shows item difficulty?
a) The proportion of students who chose each distractor
b) The proportion answering correctly*
c) The point-biserial correlation
          

Reformat the MCQs below into this exact plain-text layout (single correct answer per item):

Rules:
• One blank line between items.
• First line: “Q#. [stem]”.
• Options on their own lines, labelled a), b), c), … .
• Put a single * at the end of the correct option line.
• Keep the original wording; no explanations, feedback, or extra text.
• Use straight quotes and standard punctuation.

Return ONLY the reformatted items.

[PASTE QUESTIONS HERE]

How to import a QTI file into Canvas:

1. Go to your Canvas course.
2. Select Settings → Import Course Content.
3. Choose QTI .zip file as the content type.
4. Upload and click Import.

You can read this comprehensive Canvas page for more guidance on importing quizzes on Canvas through QTI packages.

Step 3 – Add brief feedback to each option

AI can get you to a good set of MCQs quickly, but learning improves when students get immediate, targeted feedback—even (especially!) on wrong answers. Short, option-level comments help students correct misconceptions, build confidence, and remember why an answer works.

Provide concise, option-level feedback for the following multiple-choice question(s): [paste question(s)].

Write 1–2 sentences of feedback for each option (a, b, c, …):
• Say why the option is correct or incorrect.
• Keep feedback specific to this question (no generic tips).
• For incorrect options, explain the misconception without revealing the correct answer.
• Use student-friendly, supportive language.
• If the item is ambiguous or more than one option seems correct, briefly flag the issue.

How to add option-level feedback in Canvas

1. Open your course → Quizzes → click the quiz name.
2. Click Edit, then go to the Questions tab.
3. Find the question you want and click the pencil icon.
4. In Answers, use the ⋯ menu under each choice to open the feedback box.
5. Enter a brief note for that choice (you can use the rich editor for text, links, or media).
6. Click Done, then Save.

When using these prompts to create MCQs with Copilot, please keep in mind that:

• Using a GenAI tool should be understood as an ongoing collaboration between you and the tool: iterative prompting and continuous evaluation of output will be needed to refine the questions.

• CoPilot will produce the best results when you use a prompt to generate the question(s) first and, once these are fully refined, a different one to generate feedback for the answers.

• Consider how to adapt the pre-made prompts to meet your specific requirements and preferences. For example, you might want to provide more detailed information about the learning outcomes or include a sample MCQ.

• When evaluating the CoPilot-generated MCQs, keep in mind that the tool’s main weaknesses are its limitations to provide high-quality distractors (i.e., plausible incorrect answers that “compete” with the correct answer) and the occasional generation of content-related inaccuracies.

• And, last but not least, do not underestimate the importance of your human intervention: it will be essential to ensure that the final MCQs are accurate, reliable, unbiased, and above all, appropriately aligned with the learning outcomes your students are working towards.