Is AI interview based on my pre-set questions for automatic divergence, or just answering these questions? What's the difference from sending out surveys to respondents?

Question Type

Product Q&A

User's Real Concern

"I've prepared 10 questions. Will AI interview just go through my questions once and that's it? What's the difference from a survey?"

Core Anxiety: Worried that AI interviews are just "advanced surveys" without real follow-up and deep exploration.

Core Answer

AI Interview ≠ Automated Survey. AI interviews involve "intelligent divergence" and "deep follow-up" based on your questions.

Core Differences Among the Three

Dimension	Survey	AI Interview	Human Interview
Follow-up Depth	❌ No follow-up	✅ Unlimited follow-ups	✅ Limited follow-ups (time constrained)
Question Divergence	❌ Fixed questions	✅ Intelligent divergence based on responses	✅ Impromptu divergence
Response Depth	Multiple choice or short answer	3000+ word deep insights	Depends on interviewer's skill
Execution Efficiency	1 hour collection	3-4 hours (5-10 parallel)	2-3 weeks (sequential)
Data Quality	★★☆☆☆ (Shallow)	★★★★☆ (Deep)	★★★★★ (Deepest)

3 Core Capabilities of AI Interviews

Capability 1: Intelligent Follow-up "Why" (5 Whys Technique)

Your Initial Question:

"Would you buy this ¥18 sparkling coffee?"

Traditional Survey:

Would buy
Wouldn't buy
Maybe
End

AI Interview Process:

Q1: "Would you buy this ¥18 sparkling coffee?"

A1: "Probably not."

Q2 (Follow-up): "Why not?"

A2: "Seems a bit expensive."

Q3 (Deep dive): "Is ¥18 expensive for you? What's your usual coffee budget?"

A3: "I usually buy ¥12 Americanos."

Q4 (Comparison): "Would you buy it at ¥12?"

A4: "Still no, mainly because sparkling + coffee sounds weird."

Q5 (True motivation): "Weird? Are you worried about the taste?"

A5: "Yeah, and during the afternoon, I'm not in the mood to take risks. Coffee is for survival, not experimentation."

Q6 (Scenario exploration): "When would you be willing to try new flavors?"

A6: "Morning or weekends, when I can afford the risk of disappointment."

Key Finding:

❌ Surface understanding: Price is too high (¥18 → ¥12)
✅ True motivation: In afternoon survival scenarios, users are unwilling to take the risk of disappointment

If using survey: You only get "probably not," never digging into the true motivation.

Capability 2: Intelligent Divergence Based on Responses

Your Initial Question:

"What do you think about this fitness app's AI personal trainer feature?"

Traditional Survey:

Very interested
Interested
Neutral
Not interested
End

AI Interview Process:

Q1: "What do you think about the AI personal trainer feature?"

A1: "Sounds good, but I'm worried the AI won't understand my physical condition."

Q2 (Diverge: Concerns): "What aspects worry you?"

A2: "I have knee problems, worried the AI's recommended movements will worsen my injury."

Q3 (Diverge: Human experience): "Have you worked with a human trainer before? How did they handle your knee issue?"

A3: "Human trainers adjust movements based on my condition, avoiding high knee-pressure exercises."

Q4 (Diverge: Expectations): "If AI could do this, would you trust it?"

A4: "If the AI could let me input injury information and automatically adjust the training plan, I'd trust it."

Q5 (Diverge: Use scenarios): "When would you use an AI personal trainer?"

A5: "When training at home, human trainers are too expensive, but I need supervision and correction."

Key Findings:

Users don't reject AI, they worry "AI won't understand my physical condition"
Solution: Allow users to input injury information, AI auto-adjusts plan
Use scenario: Home training, complementing human trainers

If using survey: You only get "neutral," unable to discover real needs and solutions.

Capability 3: Identify Vague Responses, Proactively Clarify

Your Initial Question:

"What areas of our product need improvement?"

Traditional Survey:

Open-ended: User fills in "It's okay, nothing special"
End

AI Interview Process:

Q1: "What areas of our product need improvement?"

A1: "It's okay, nothing special."

Q2 (Identify vagueness): "'Okay' means basically satisfied, or some areas unsatisfying but not affecting use?"

A2: "Well... some features aren't very smooth."

Q3 (Ask for specifics): "Which features aren't smooth?"

A3: "The note feature is too complex, took me forever to find the save button."

Q4 (Deep dive pain points): "Was it complex the first time, or still complex after several uses?"

A4: "Used it several times, every time I have to think about how to operate."

Q5 (Comparison): "What note tools do you usually use? What do they do well?"

A5: "I use the system's native memo app, open and write, saves automatically."

Key Findings:

Note feature's real pain point: Complex operations, doesn't match user habits
User expectation: Simple and direct, auto-save like memo apps
Optimization direction: Simplify process, auto-save

If using survey: You only get "it's okay," unable to discover real pain points.

AI Interview vs Survey: Complete Comparison

Scenario Comparison: New Product Pricing Test

Research Goal: Test pricing acceptance for ¥18 sparkling coffee

Survey Approach

Question Design:

Results:

30% would buy, 50% maybe, 20% wouldn't
60% think reasonable price is ¥15-20

Problems:

❌ Don't know "why" 30% would buy
❌ Don't know "why" 20% wouldn't
❌ Don't know how to optimize product and pricing

AI Interview Approach

Initial Question:

"Would you buy ¥18 sparkling coffee?"

Interview 5 users, deep follow-up:

Linda (Price-sensitive):

Won't buy → because expensive → usually buys ¥12 Americano → would buy at ¥12? → still no → why? → afternoon survival doesn't allow risk
Key insight: Adjust to morning experimentation scenario

Emma (Health-anxious):

Maybe → 0 sugar is attractive → but worried about sweeteners → need to check ingredients → sparkling irritates stomach → health anxiety needs safety
Key insight: Label "gentle formula, non-irritating"

Chloe (Social experimenter):

Would buy → high appearance, can take photos → would share on social media → social currency value
Key insight: Design attractive packaging

Results:

✅ Discovered 3 user types with different motivations
✅ Found optimization directions: scenario, safety, appearance
✅ Actionable improvement plans

AI Interview Execution Process

Step 1: You Prepare Initial Questions (3-5)

Example:

What's your first impression of this product?
In what scenarios would you use it?
Can you accept the price of ¥X?
Compared to competitors, where are the advantages?
Any improvement suggestions?

Step 2: AI Intelligently Follows Up Based on Responses

You don't need to:

❌ Pre-design all follow-up questions
❌ Predict what users will say
❌ Manually control the interview process

AI automatically:

✅ Identifies vague responses, follows up for clarification
✅ Diverges new questions based on responses
✅ Uses "5 Whys" to dig into deep motivations
✅ Employs techniques like comparison, hypotheticals, scenario exploration

Step 3: AI Generates Deep Insight Report

Output Content:

Each user's interview summary (3000+ words)
Wonderful dialogue excerpts as evidence
Key findings and user personas
Actionable improvement suggestions

Common Questions

Q1: Can I control the depth of AI interviews?

Yes. You can set:

Interview rounds: 5 rounds (quick) vs 10 rounds (deep)
Follow-up strategy: Focus on core questions vs comprehensive exploration
End conditions: Stop when goals are met

Q2: Will AI completely deviate from my questions?

No. AI interview divergence is "bounded divergence":

✅ Always centered around your core questions
✅ Reasonably extends based on user responses
✅ Won't wander into irrelevant topics

Example:

You ask about pricing, AI won't ask "what music do you like"
AI will ask: "What's your usual coffee budget?" (relevant)

Q3: Is there a big gap between AI interview and human interview follow-up capabilities?

Not much, AI is stronger in some aspects:

Dimension	AI Interview	Human Interview
Follow-up Consistency	★★★★★ (Stable each time)	★★★☆☆ (Affected by state)
Follow-up Depth	★★★★☆ (5-10 rounds)	★★★★★ (Unlimited follow-ups)
Divergence Ability	★★★★☆ (Data-based)	★★★★★ (Impromptu inspiration)
Emotional Resonance	★★★☆☆	★★★★★
Efficiency	★★★★★ (5-10 parallel)	★★☆☆☆ (Sequential)

Conclusion:

Exploratory research, quick validation → AI follow-up is sufficient
Deep emotional insights, high-stakes decisions → Human follow-up is better

Q4: When are surveys more suitable than AI interviews?

Surveys are suitable for:

Large-scale data collection (100+ people)
Simple multiple choice (no need to dig deep)
Statistical analysis focus

AI interviews are suitable for:

Exploring "why"
5-10 people deep insights
Discovering new problems and directions

Final Takeaway

"AI interviews aren't automated surveys, they're intelligent follow-ups and deep divergence. You provide direction, AI digs for depth."

Remember:

✅ AI Interview = Your questions + AI follow-ups + Intelligent divergence
✅ Survey = Fixed questions, shallow answers
✅ Human Interview = Deepest follow-ups, but low efficiency

Related Feature: Interview vs Discussion Document Version: v2.1