The big idea: AI can hand you a hundred questions in a minute, but only your team can choose the fifteen that will get a real person to tell you a real story.
In 60 seconds
- Draft your own questions first, then use AI to stretch them, not write them.
- Cut any question that is leading, double-barreled, yes/no or full of jargon.
- Ask for stories about specific past moments, and order your questions as a conversation arc.
Do this: submit Deliverable #2: 15 open-ended interview questions plus a 5 to 6 sentence note on how AI helped.
How to build questions with AI and keep them human
Your interviews can only be as good as your questions. In Deliverable #2 your team builds 15 open-ended questions with AI as a thinking partner, not the final author. You draft first, let AI stretch your thinking, then make every question sound like a real person asking it.
Draft first
Write five to seven questions yourselves, before any AI. Ask: what do we want to understand?
Human: your curiosity sets the direction.
Expand with AI
Ask AI for more ideas, better wording, more open versions and versions for other stakeholder groups.
AI: keep asking it why, and push back.
Curate and humanize
Pick the best 15. Rewrite each so it sounds natural out loud and works for more than one kind of person.
Human: cut, combine, rewrite.
Reflect
Write a 5 to 6 sentence note: how AI helped, what you changed and which question is strongest, and why.
Bring: one printed page to class.
What your questions should uncover
Aim for questions that reveal
- Needs. What people are trying to get done.
- Frustrations and barriers. What gets in the way, and what they do about it.
- Motivations. Why they do it the way they do.
- Success stories. When it went right, and what made the difference.
- Opportunities and hopes. What a better future would look like to them.
Openers that get stories
- Tell me about the last time you… A specific moment beats a general opinion.
- Walk me through how you… Steps reveal the workarounds.
- What is the hardest part of… Pain points, in their words.
- When things do not work, what do you do instead? Hacks point to unmet needs.
- If you could wave a magic wand… Hopes and ideals.
Four flaws that cost you a good answer
Leading
“Wouldn’t you agree students bring fresh ideas?” It tells people what you hope to hear, and polite people agree.
Fix: “What, if anything, do students bring that is different?”
Double-barreled
“What do you value in a partnership and how often do you hire students?” Two questions, one answer, and you cannot tell which half it covers.
Fix: split it and ask the more important half first.
Yes/no
“Do you like working with interns?” One word ends the conversation before the story starts.
Fix: “Tell me about the last intern you worked with.”
Jargon
“How can we leverage synergies with stakeholders?” Business-school words can confuse people or make them guess what you want.
Fix: use the words the person would use at work or at home.
AI is a thinking partner
AI is fast at volume and at rewording. It does not know your challenge, your stakeholders or how a question lands in a real conversation. Your team decides what stays, what changes and why.
Write for more than one person
Your 15 should work for students, faculty, small businesses, large employers and community leaders. If only one group can answer a question, add a neutral version or save it as a follow-up.
Build a conversation arcOrder your questions so the interview flows
A list of 15 questions is not a script. The Stanford d.school empathy field guide describes an interview as a conversation with a shape. Here is that shape in short:
- Introduce yourself and the project. Simple and friendly. Say why you are there.
- Build rapport. Start with easy, non-threatening questions about their role or their day.
- Evoke stories. Ask about specific past moments: the first time, the last time, the best or worst time.
- Explore emotions. “How did that feel?” “What was going through your mind?”
- Question statements. Gently probe what they said: “You said it was easy. What made it easy?”
- Thank and wrap up. Ask if there is anything you should have asked, and who else you should talk to.
Focus on the past and present, not the future. People describe what they did far more accurately than what they would do.
Question themes from the GLS inspiration guideThirteen angles so your list is not 15 versions of one question
The Smith GLS guide to inspiration-phase questions (built from IDEO’s approach) sorts good questions into themes. Use them as a coverage check:
| Theme | Example: campus dining wait times |
|---|---|
| Personal stories | Tell me about the last time you grabbed lunch between classes. |
| Routines and context | Walk me through a typical Tuesday from your 11:00 class to your 1:00 class. |
| Emotions and values | How did you feel when the line was out the door? |
| Needs and motivations | What matters most to you when you pick where to eat on campus? |
| Workarounds and hacks | When the line is too long, what do you do instead? |
| Pain points | What is the hardest part of eating on campus on a busy day? |
| Dreams and ideals | If you could design the perfect lunch break, what would it look like? |
| Social influences | Who do you usually eat with, and how do they shape where you go? |
| Comparisons | How does lunch here compare to other places you have eaten regularly? |
| Extreme users | (To a dining worker) What do you notice about the rush that students never see? |
| Behaviors and journeys | How do you decide where to eat on a given day? |
| Delight and surprise | Tell me about a time eating on campus surprised you in a good way. |
| Playful reframing | If the dining hall were a person, how would it treat you at noon? |
Prompts that keep your team in chargeHow to use AI without letting it write the list
- Expand: “Here are our seven questions about [your team’s challenge]. Suggest ten more that would uncover barriers and workarounds.”
- Open up: “Rewrite these questions to be more open-ended and curiosity-driven. Explain each change.”
- Tailor: “Adapt question 4 for a student, a small business owner and a university leader.”
- Critique: “Which of these questions are leading, double-barreled, yes/no or full of jargon? Why?”
- Ask why: “Why is that version better? What would a person likely say to each one?”
- Rehearse: ask AI to role-play a busy stakeholder so you can practice follow-ups. The course AI policy lists this as an appropriate use.
Save your prompts. Your reflection note needs to say how AI helped and what you changed, and later courses ask for the prompts you used.
Follow-up probes to keep in your pocketThe best insight usually comes after the second question
Go deeper
“Why do you say that?” “Tell me more.” “What happened next?” “Can you give me an example?”
Stay neutral
Do not suggest an answer when they pause. Silence often brings the most honest part of the story.
Ask about feelings
“What were you feeling at that point?” Emotions point to what matters.
Watch, not just listen
Notice when what people say and what they do do not match. That gap is worth a follow-up.
Test the list out loudThe fastest way to find awkward wording
The dscout guide to generative research recommends reading your script out loud several times and practicing it on others to catch awkward language and jargon. Try it with a teammate:
- Read each question as if you were asking a nurse, a store owner or a first-year student. Does it still make sense?
- Time a practice run. If 15 questions take 10 minutes, the answers are too short.
- Mark any question you stumbled over. Rewrite it in fewer words.
Privacy and integrityWhat stays out of public AI tools
- Do not paste names, contact details or private information about stakeholders into a public AI tool. Describe roles instead (“a small business owner”).
- Do not submit AI questions without review and edits. The rubric gives zero process points for lists clearly copied from AI without thought.
- Be clear in your note about what AI did and what your team did.
Try it: Interview Question Checker
Tool · BUAD 317
Interview Question Checker
Catch leading, double-barreled, yes/no and jargon questions before you ask them.
- 1Paste
- 2Fix
- 3Share
Ready to ask
What your list helps uncover
All questions
Put it to work: Deliverable #2 AI+Human Interview Questions
This is Deliverable #2, AI+Human Interview Questions, a team assignment worth 10 points. Upload your list to Canvas and bring one printed page to class: 15 clear, open-ended questions that work across stakeholders for the design challenge, plus a 5 to 6 sentence note on your process. You will refine the list together in class, then use it for your Deliverable #3 stakeholder interviews.
Build your list in six steps
Draft without AI
Each teammate writes questions alone, then the team keeps the five to seven that get at what you most want to understand.
Your deliverable: a human-only first draft. Save it so you can show what changed.
Expand and improve with AI
Generate more ideas, better wording and versions for different stakeholders. Ask why, push back and save your prompts.
Your deliverable: a long list of candidates, with notes on what AI suggested.
Run the checker
Paste every candidate into the Interview Question Checker. Fix or cut each red flag.
Your deliverable: a clean list with no leading, double-barreled, yes/no or jargon questions.
Curate the best 15
Cover needs, frustrations, motivations, success stories, barriers, opportunities and hopes. Order them as a conversation arc.
Your deliverable: 15 questions that a student, a business owner and a university leader could all answer.
Say them out loud
Read the list to a teammate playing a stakeholder. Rewrite anything that sounds stiff.
Your deliverable: questions that sound natural in conversation.
Write the note
In 5 to 6 sentences: how AI helped, what you changed or improved, and which question is strongest and why.
Your deliverable: one printed, clearly formatted page, also uploaded to Canvas on time.
Example: one question, AI to humanHow a question changes across the process
| Stage | Question | What changed |
|---|---|---|
| Human draft | Is onboarding good here? | Yes/no and vague, but it names what we care about. |
| AI version | How effectively does the onboarding program enable new hires to achieve productivity, and what improvements would you suggest? | Open, but double-barreled and full of jargon. |
| Final | Tell me about your first week here. Who showed you how things really work? | A specific story, plain words, and a follow-up that gets at the hidden training. |
Sample note: “We drafted seven questions before using AI. AI helped us add questions about barriers and tailor wording for students and employers. It also made several questions longer and more formal, so we cut jargon and split two double-barreled questions. We replaced three yes/no questions with ‘Tell me about the last time’ openers. Our strongest question is number 5, because it asks for a specific experience instead of an opinion.”
What the rubric rewardsWhere the 10 points come from
Quality of questions: 4
Consistently open-ended, clear, conversational and insight-driven. They invite stories, experiences and emotions, not short answers.
Alignment with the challenge: 2
Questions explore needs, barriers, opportunities and perspectives from multiple sides of the design challenge.
AI + human process: 2
Your note clearly explains how AI was used, what you changed and why.
Variety and professionalism: 1 + 1
Questions work for multiple groups. The page is organized, on time, uploaded and printed.
Do this, not thatHabits that make a list interview-ready
Do this
- Draft your own questions before opening an AI tool.
- Ask about specific past moments.
- Keep one idea per question.
- Use the words your stakeholders use.
- Keep follow-up probes ready.
Not that
- Pasting the challenge into AI and printing what comes back.
- Asking people to predict whether they would use your idea.
- Writing 15 questions only a business owner could answer.
- Hinting at the answer you hope for.
- Putting names or private details into a public AI tool.
Before you submitCheck these before you print
- [ ] We have exactly 15 questions.
- [ ] Every question is open-ended. None can be answered with yes or no.
- [ ] No question is leading, double-barreled or full of jargon.
- [ ] The list covers needs, frustrations, motivations, success stories, barriers, opportunities and hopes.
- [ ] The questions work for more than one stakeholder group.
- [ ] We read the list out loud and it sounds like us.
- [ ] Our 5 to 6 sentence note says how AI helped, what we changed and which question is strongest, and why.
- [ ] The page is uploaded to Canvas and printed for class.
Go deeper
Back to all BUAD 317 topics · Next: Stakeholder Interviews, Consent and Ethics
In-class team activity · 25 min
Question Crash Test
By the end: Five interview questions for your team's challenge, drafted by hand, stretched with AI, crash-tested on a real person in the room and ready for your Deliverable #2 list. Your instructor will give you the password in class.
Part of the Smith GLS human-centered leadership sequence: IB 407, BUAD 317 and BUAD 497, toward the Human Centered Consulting Certificate.
Frequently asked questions
How do you write interview questions with AI?
Use AI as a thinking partner, not the final author. Draft five to seven questions as a team before using any AI, then ask AI for more ideas, better wording and versions for other stakeholder groups. Curate the best 15, rewrite each so it sounds natural out loud, and write a short note on how AI helped and what your team changed.
What makes a bad interview question?
Four common flaws cost you good answers: leading questions, double-barreled questions, yes/no questions and jargon. A leading question tells people what you hope to hear. A double-barreled question asks two things at once. A yes/no question ends the story before it starts. Jargon confuses people. Fix them with neutral, single, open questions in the words people use at work or at home.
What are good open-ended questions to start a user interview?
Strong openers invite stories instead of opinions. Try "Tell me about the last time you…," "Walk me through how you…," "What is the hardest part of…," "When things do not work, what do you do instead?" and "If you could wave a magic wand…." Specific moments, steps, pain points, workarounds and hopes reveal far more than general opinions.
How should you order interview questions?
Order your questions as a conversation arc, not a list. Following the Stanford d.school empathy field guide, introduce yourself and the project, build rapport with easy questions, evoke stories about specific moments, explore emotions, gently probe what people said, then thank them and ask who else you should talk to. Focus on the past and present, since people describe what they did more accurately than what they would do.
What is the AI+Human Interview Questions assignment in BUAD 317?
Deliverable #2, AI+Human Interview Questions, is a team assignment worth 10 points. Teams upload 15 clear, open-ended questions that work across stakeholders, plus a 5 to 6 sentence note on how AI helped and what the team changed, and bring one printed page to class. The rubric gives zero process points for lists clearly copied from AI without thought.