Interaction Design in the Age of AI

Revisions

What changed after your Oct 5 comments

Revised Oct 18, 2026

Each row is one of your comments and where I fixed it.

D2 · Screener

What six screener respondents said

Sep 16, 2026 · revised Oct 5, 2026

I sent a six-question screener on Typeform on Aug 20, 2026, back when the project was framed as career coaching, and got six responses. I numbered the respondents P1 to P6 in the order Typeform lists them. All answers below are exactly how they wrote them, including spelling. I left out email addresses.

Full responses, by question

Q1 · What year are you in school, and what are you currently job/internship hunting for (if applicable)?

Q2 · Think of a career or skill goal you set for yourself recently (job search, learning a skill, networking, etc). What was it?

Q3 · What made it hard to actually follow through on that goal, day to day?

Q4 · Have you ever used an app, planner, or AI tool to help you stick to a goal like this? What happened?

Q5 · If a tool could break a big goal into small steps and check in with you, would that help? Why or why not?

Q6 · Would you be open to a 15-min video chat about this?

How I coded the responses

Each column comes from one question above. The Q3 column is my own short label for each answer. Everything else is taken straight from the answers.

What stood out

Five of six talked about keeping their effort going. In Q3, P1 said "willpower," P2 "busy school works," P3 "consistency," P5 "some days I’m unmotivated," and P6 "nobody keeps me accountable, motivation issues." Three also didn't know what to do next: P1 ("lack of expert guidance"), P5 ("learn more specifics"), and P6 ("not sure where to start"). I can't really tell what P4 meant by "the greed" from two words, and P4 didn't want a follow-up.

Most people had used a tool, but few said how it went. Five of six said yes to Q4. Q4 also asked "What happened?", but only P2 ("worked efficently") and P6 ("couldn't keep up") said how it went. P6 also guessed why: "maybe because my plan was not strategic enough." The others just named the tool. I couldn't ask follow-ups in a survey, so I added this to the interview guide (D3, Q4).

The two people who said no to Q5 still had trouble with their goals. Q5 asked if a tool that breaks a goal into small steps and checks in would help. P1, P2, P4, and P6 said yes. P3 said no because they don't use apps and one didn't help before. P5 said no because "I like to plan my own things out," but in Q3 still said some days they're unmotivated. With only six responses I'm not calling this a pattern. I just took it as something to ask about in the interviews.

What the screener couldn't answer

Each row is something the screener didn't answer, and the interview question (D3) I wrote to cover it.

D3 · Interviews

Interviews with P1, P2, and P6

Guide revised Oct 5, 2026 · interviews Oct 12–17, 2026

I rewrote the interview guide so the questions are open and don't lead people, and so they cover the gaps above. Then I interviewed three of the screener respondents with it. They keep the same numbers as in the screener. Every question and answer is below.

P1 and P2 said yes to a follow-up in the screener. P6 said no at first, but I needed one more interview, so I asked P6 again in October and they agreed. All three were okay with me taking notes. The answers below are from my notes. I wrote them down close to what each person said, but they aren't word for word.

Interview guide

Text in [brackets] comes from that person's screener answers. I asked follow-ups one at a time, after they answered the main question. I only asked probes if the answer didn't already cover them. Before Q5, I described the tool the same way the screener did: "a tool that breaks a big goal into small steps and checks in with you."

1. In the screener, you described a goal: [Q2 answer]. Can you tell me about it?
   Probe: Why did you choose that goal?
2. What, if anything, have you done toward that goal since you filled out the screener?
   Probe: How, if at all, has that changed over time? If they've done nothing: Why not?
3. You said [Q3 answer] made it hard to follow through. Can you tell me about a time that happened?
   Probe: Why do you think that happens?
4. You mentioned [Q4 answer]. How did you use it?
   Follow-ups: Are you still using it? Why or why not? If they used no tool: How do you keep track of the goal, if at all?
5. How, if at all, would you want a tool like this to work for you?
   Follow-up: Here are a few ways it could work.
   • A: You set the steps and schedule; the tool only reminds you.
   • B: The tool suggests steps and a schedule; you approve or change them.
   • C: The tool sets the steps and schedule; you follow them.
   • D: None of these.
   Which, if any, is closest to what you would want? Why?
6. Do you think you would use a tool like this? Why or why not?
   Follow-up, if yes: Do you think you would continue to use it? Why or why not?
7. Is there anything about working toward this goal that I didn't ask about and you think matters?

Q5 and Q6 ask about a tool that doesn't exist yet. So I count those answers as weaker evidence than Q2 to Q4, which are about what people actually did.

Full interviews

Each question is written the way I asked it, with the brackets filled in from that person's screener. I only listed the probes and follow-ups I actually asked. My own notes are separate, at the end of each interview. Highlighted parts are the ones I used in the synthesis (D4).

P1

Oct 12, 2026 · in person · 30 min · consented to notes

My notes after the interview (my take, not P1's words). What surprised me: P1 had plenty of drive. Their motivation dropped once they got stuck on the math and had no one to ask. Compared with the screener: This matches what P1 wrote. ChatGPT was good for planning out a syllabus, but it couldn't help when something broke. Follow-up I should have asked: "When ChatGPT generated your daily schedule, what did you do on the very first day you missed a task?"

P2

Oct 16, 2026 · in person · 30 min · consented to notes

My notes after the interview (my take, not P2's words). What surprised me: P2 really wants to stay in control. They don't want complicated features or an AI planning their day. Compared with the screener: This matches "busy school works," since school deadlines push out personal projects. The tool they called "efficient" turned out to be simple time blocking. Follow-up I should have asked: "What is the trigger that actually makes you check off a task on a busy day?"

P6

Oct 17, 2026 · in person · 30 min · consented to notes

My notes after the interview (my take, not P6's words). What surprised me: Organizing was basically a way for P6 to put off the real work, because actually making something felt harder. Compared with the screener: This matches "not strategic enough" and the accountability problem. The planner stopped working because the lists kept getting too big. Follow-up I should have asked: "Would an AI check-in feel like real accountability to you, or does accountability have to come from another human?"

What changed from the first guide

D4 · Synthesis

What the screener and interviews add up to

Oct 18, 2026 · from the Oct 12–17 interviews

This part only uses the six screener responses (D2) and the three interviews (D3). Every finding lists the answers it came from, by participant number and question.

Affinity map

I grouped the interview answers by what they were about. Each row is one group with the answers in it. Screener answers are marked "Screener."

Journey: from setting a goal to dropping it

These stages come from all three interviews, but not everyone talked about every stage. Each cell says who said it.

Where participants disagreed

How much should the tool plan?

• P1 (picked B): wants it "to plan the path since it knows more than me."
• P6 (picked B): wants "something to make the first move and cut it down for me."
• P2 (picked A): only wants reminders. "I don't want AI planning my day."

P2 is also the only one whose current tool still works every day (Q4 follow-up), so maybe a planning tool matters less for P2.

P2's answers don't fully match.

• P2 picked A, where the tool only reminds you (Q5 follow-up).
• But when I asked openly how it should work, P2 said it could "look at my calendar, find the free time between classes" and suggest one small task (Q5). That's closer to B, because the tool is choosing when and what.

So I think P2 doesn't mind suggestions, as long as they're small and fit into time P2 already has. P2 seems to push back on the word "planning" more than on the tool actually helping. I didn't catch this during the interview, so it's something to ask about in usability testing (D7).

Check-ins.

• P6 asked for a check-in (Q5).
• P1 and P2 both said constant notifications would make them stop (Q7; Q6 follow-up).

Maybe one check-in at a time the person picks could work for both, but I didn't test that.

Limits

• Three interviews is a small sample.
• All three said yes to the tool question in the screener (Q5), so I didn't hear from the people who doubted a tool the most. P3 ("I don't use apps for anything") and P5 ("I like to plan my own things out") weren't interviewed.
• Q5 and Q6 ask about a tool that doesn't exist, so those answers are weaker than what people said they actually did in Q2 to Q4.
• All three are students or job seekers in design or computing.

What this means for D5

1. The tool suggests steps, and the person can approve, change, or skip them. The tool never sets the plan by itself (no one picked C in Q5).
2. The tool shows one next step at a time, sized to the time available (P2 and P6, Q5).
3. After a missed day, the tool re-plans instead of piling up overdue tasks, and it doesn't make the person feel guilty (P1, Q6; P6, Q4 and Q6 follow-up).
4. The tool remembers where the person got stuck and picks up from there (P1, Q6 follow-up).
5. The tool only checks in a few times, at a time the person picks (P1, Q7; P2, Q6 follow-up; P6, Q5).
6. The tool connects to a calendar people already use, so they don't have to manage a new app (P2, Q6 and Q4 follow-up).

Open questions for D5

• Would a check-in from AI feel like real accountability, or does it need to come from a person? (from my P6 notes)
• What do people do on the first day they miss a step? (from my P1 notes)
• Can the tool help with a problem in a specific subject, like P1's math, or only with scheduling?