The big idea: AI can find what is already known and draft questions fast, but only your judgment turns its gaps into clear questions a real person wants to answer.
In 60 seconds
- Deliverable #2 is individual: bring a set of AI-generated interview questions to Class 4 to discuss.
- Follow the five-step framework and the four guidelines: transparency, bias awareness, inclusion and collaboration.
- Clear questions ask one thing, invite stories, cover several question types and trace back to a source.
Do this: generate your Deliverable #2 questions with AI, then vet, rewrite and disclose them before Class 4.
How to get interview questions from AI and keep people at the center
For Deliverable #2, each of you, on your own, brings a set of AI-generated interview questions about the class design challenge to Class 4. We discuss them in class, and your interview team then builds its conversation guide from the best of them. The goal is not to let AI write your interviews. It is to use AI responsibly for secondary research, so your questions start from what is already known and point at what is not.
AI is a partner in curiosity, not a shortcut
The BUAD 317 framework treats AI as a way to broaden and deepen secondary research while you keep people, with their needs, wants, values and contexts, at the center. You stay responsible for every question you bring.
Four guidelines for every AI session
Transparency
Cite your sources and say when you used AI summaries.
Do it: keep the tool name and your prompts, and add the AI disclosure statement.
Bias awareness
Ask whose voices the AI output amplifies and whose are missing.
Do it: write one sentence on who the sources left out.
Inclusion
Actively seek diverse and marginalized perspectives.
Do it: make sure your questions work for more than one group.
Collaboration
Use AI to augment human judgment, empathy and creativity, never to replace them.
Do it: rewrite the weakest AI questions in your own words.
The five-step framework: from a challenge to interview questions
Scope the question
Start from the How Might We challenge. Ask AI to reframe it into focused queries about people’s needs, contexts and experiences.
Example output: research queries from academic, industry and community angles.
Search and gather
Use AI to search academic work, industry reports, case studies and the open web. Look past the dominant, institutional voices.
Example output: a curated, annotated source list.
Synthesize
Cluster, summarize and compare findings. Turn them into insights, empathy maps or proto-personas that show the people behind the data.
Example output: themes, tensions and open questions.
Communicate
Turn research into plain-language stories or visuals that people can engage with. Keep the human experience at the center.
Example output: a short presentation or infographic.
Find the gaps
Ask AI for contradictions, gaps and unanswered questions. Those gaps become what you ask real people.
Your D#2: interview questions that reach into the gaps.
The Perplexity workflow, literature review steps and annotated bibliography guide live on the Class 2 secondary research page. This page picks up at step 5.
Prompt for all five question types
AI tools default to opinion questions (“How satisfied are you…?”). Ask for each type on purpose, using the five types from Casual Learning Conversations.
Prompts that keep you in charge
- Ground it. “Based on the sources you found, what do we still not know about how [group] experiences [challenge]?”
- Ask by type. “Write three discovery and three behavior questions that ask for stories about past experiences.”
- Use TEDW. “Rewrite these so each starts with Tell me about, Explain, Describe or Walk me through.”
- Widen the circle. “Whose perspective is missing from these questions? Add questions for them.”
- Critique. “Which of these are leading, double-barreled, yes or no, or full of jargon? Why?”
Then do the human part
- Cut. Drop anything you would feel silly asking a stranger.
- Rewrite. Put the weakest questions in your own words.
- Trace. Each question should connect to something you learned from a source.
- Say it out loud. If it sounds like a survey, make it sound like you.
- Park solutions. “Would you use…” questions wait for prototype feedback in Part D.
Four flaws to fix in AI questions
Leading
“Wouldn’t a faster online portal make it better?” It tells people what you hope to hear, and polite people agree.
Fix: “Tell me about the last time you checked on your application.”
Double-barreled
“How satisfied are you with the service and how could it improve?” Two questions, one answer.
Fix: split it and ask the more important half first.
Yes or no
“Is the office staff helpful?” One word ends the conversation before the story starts.
Fix: “Describe the last time you asked the staff for help.”
Jargon
“How can the city leverage digital touchpoints?” AI loves these words. People do not use them.
Fix: use the words people use at home or at work.
AI use and disclosure in BUAD 317What is appropriate, and the statement to include
The course AI policy encourages AI as a catalyst for creativity and human-centered leadership, used with integrity, transparency and discernment. You take full responsibility for the accuracy, originality and ethics of what you submit.
- Appropriate: brainstorming, finding and summarizing sources you then read, and practicing interview questions with an AI role-play and reflecting on how you did.
- Not appropriate: submitting AI-generated work without review, edits or acknowledgment.
- Privacy: never paste names, contact details or private stories from real people into a public AI tool. Describe roles instead.
Disclose any AI use and how you used it (initial research, outlining, editing). The syllabus gives this statement to adapt:
“The author(s) would like to acknowledge the use of [Generative AI Tool Name], a language model developed by [Generative AI Tool Provider], in the preparation of this assignment. The [Generative AI Tool Name] was used in the following way(s) in this assignment [e.g., brainstorming, grammatical correction, citation, which portion of the assignment].”
For keeping an AI log across the senior project, see Using AI in Consulting Research, the Right Way (BUAD 497).
Try it: AI Question Vetting Board
Tool · BUAD 317
AI Question Vetting Board
Paste the interview questions AI gave you. See their type, clarity and the human edits still needed before Class 4.
- 1Paste
- 2Vet
- 3Bring
Question clarity, self-check
Question by question
All checks
Put it to work: Deliverable #2 AI-generated interview questions
Deliverable #2, AI-generated interview questions, is individual. Bring your set to Class 4, ready to discuss. Your instructors will confirm how many questions to bring. In class you compare sets, and your feet on the street interview team builds its guide for Deliverable #3 Part A from the strongest questions.
Build your set in six steps
Scope with AI
Paste the class design challenge and ask AI to reframe it into focused research queries from academic, industry and community angles.
Your deliverable: three to five research queries you kept, and the prompt you used.
Gather and read sources
Have AI find sources, then open them. Check that each one says what the summary claims.
Your deliverable: a short list of sources you checked yourself.
Ask for the gaps
Ask AI what the sources disagree on, what they leave out and whose voices are missing.
Your deliverable: a few gaps worth asking real people about.
Generate questions by type
Prompt for screening, solution, discovery, behavior and opinion questions that reach into those gaps. Ask for TEDW openers.
Your deliverable: a raw AI list, saved as it came out.
Vet and humanize
Run the list through the AI Question Vetting Board. Fix the flags, rewrite the weakest questions yourself and mark them, and note whose voices are missing.
Your deliverable: your edited set, each question clear and tied to what you learned.
Disclose and bring it
Add the AI tool, your prompts and the disclosure statement. Read every question out loud once.
Your deliverable: your Deliverable #2 set, ready to discuss in Class 4.
What the Question Clarity rubric looks forFour levels, from the BUAD 317 syllabus
| Level | What it looks like |
|---|---|
| Excellent | Questions are clear and concise, and they summarize, evaluate and reflect on each source well. |
| Good (meets expectations) | Questions are mostly clear and concise, with a few minor issues in detail or structure. |
| Needs work | Questions are unclear, too vague, or missing needed detail. |
| Unsatisfactory | Questions are poorly written, unclear, or lack meaningful content. |
“Reflect on each source” means a reader can see where a question came from. The Vetting Board’s clarity level is a self-check, not a grade.
Example: one question, AI to humanA made-up city permit office challenge
| Stage | Question | What changed |
|---|---|---|
| Gap AI found | Reports describe the online form, but say little about what applicants do when they get stuck. | A gap worth asking real people about. |
| Raw AI question | How satisfied are you with the city’s permitting services and how could they be improved? | Opinion only, double-barreled, invites a vague answer. |
| Brought to class | When something in the process did not make sense, what did you do instead? | A behavior question about a real moment. It reaches the gap and invites a work-around story. |
Do this, not thatHabits that make AI questions interview-ready
Do this
- Read the sources AI cites before you trust them.
- Ask for every question type, on purpose.
- Keep one idea per question.
- Rewrite the weakest questions yourself.
- Save your prompts and name your tool.
Not that
- Pasting the challenge and printing whatever comes back.
- A list of satisfaction and rating questions.
- Asking people to predict whether they would use an idea.
- Questions only one kind of person could answer.
- Names or private details in a public AI tool.
Before Class 4Check these before you bring your set
- [ ] Every question is clear, concise and asks one thing.
- [ ] No question is leading, yes or no, double-barreled or full of jargon.
- [ ] My set covers several of the five question types, with most in discovery and behavior.
- [ ] Each question connects to something I learned from a source.
- [ ] I rewrote the weakest AI questions in my own words.
- [ ] I noted whose voices the AI output left out.
- [ ] I named the AI tool, kept my prompts and added the disclosure statement.
- [ ] I read every question out loud once.
Go deeper
Back to all BUAD 317 topics · Next: Feet on the Street Interviews: Consent, Ethics and Trustworthy Data
In-class team activity · 25 min
AI Question Swap Shop
By the end: A merged starter guide for your interview team, built from everyone's Deliverable #2 questions, with each question typed, cleaned up and tested on a live person. Your instructor will give you the password in class.
Part of BUAD 317, taught by Lane Morris, PhD and Jules Morris, PhD, in the Smith GLS human-centered leadership program, toward the Human Centered Consulting Certificate.
Frequently asked questions
What is Deliverable #2 in BUAD 317?
Deliverable #2 is an individual assignment. Each student brings a set of AI-generated interview questions about the class design challenge to Class 4, ready to discuss. Your instructors confirm how many to bring. The questions grow out of AI-supported secondary research, and your interview team later builds its conversation guide for Deliverable #3 Part A from the strongest ones.
How is Deliverable #2 graded?
The rubric focuses on Question Clarity. Excellent questions are clear and concise and reflect the sources behind them. Good questions are mostly clear with minor issues in detail or structure. Questions that need work are vague or missing detail. Unsatisfactory questions are poorly written or lack meaningful content. The AI Question Vetting Board gives a self-check, not a grade.
What are the four guidelines for using AI in BUAD 317 research?
Transparency: cite sources and say when you used AI summaries. Bias awareness: ask whose voices the AI output amplifies and whose are missing. Inclusion: actively seek diverse and marginalized perspectives. Collaboration: use AI to augment human judgment, empathy and creativity, not to replace them. Together they keep people at the center of AI-supported research.
Can I just paste the AI questions into my assignment?
No. The course AI policy says submitting AI-generated work without review, edits or acknowledgment is not appropriate. Read the sources AI cites, fix leading, yes or no, double-barreled and jargon-heavy questions, rewrite the weakest ones yourself, and add the AI disclosure statement naming your tool and how you used it.
Why should I avoid asking "Would you use…" questions now?
Deliverable #2 supports generative research, which explores people's experiences before any solution exists. People are poor at predicting what they would use, and polite people often say yes. Save "Would you use…" and recommendation questions for prototype feedback in Deliverable #3 Part D, when people can react to something real you have built.