Commonalities of Job Seekers Who Fail AI Interviews: 5 Behavioral Differences Between Those Who Pass and Those Who Fail [Actual Comparison]
I took an AI interview, but I failed. I don't know what went wrong.
Have you ever had that experience?
SHaiN, HireVue, MIWS...
More and more companies are introducing AI interview tools.
You just sit in front of the camera and answer questions from the screen.
But after it's over, many people feel, 'I should have been able to answer properly,' or 'I don't know what is being evaluated.'
There are many people who feel this way.
In this article, we compare the behaviors of job seekers who passed AI interviews with those who failed, and organize
what determined the pass/fail result into 5 differences.
Note: All cases are synthetic episodes created based on multiple consultation cases and job market data.
They do not refer to any specific individual.
What you will learn in this article
How 5 behaviors differ between those who pass and those who fail AI interviews
What AI is actually evaluating (eye contact, speaking speed, logical structure, etc.)
A preparation schedule you can start one week before the actual interview
Difference 1: Depth of Preparation
What those who passed did
What successful job seekers had in common was that they articulated 'why this company' for each company in advance.
Specifically:
Researched the company's business, products, and differences from competitors
Prepared to explain 'why I want to join this company' by comparing it with others
Picked out and prepared for 5 to 10 frequently asked questions in the industry
Organized their own episodes using the STAR method (Situation, Task, Action, Result)
In AI interviews, follow-up questions may also be asked.
'Could you tell me a little more about the reason?'
'What did you learn from that experience?'
If your preparation is thin, you will get stuck on these follow-up questions.
What rejected candidates were doing
On the other hand, many job seekers who were rejected memorized model answers by heart.
"I memorized the ideal answers generated by ChatGPT"
"I read through collections of interview answers found online"
They did the preparation, but it didn't become "their own story."
When probed further, they ran out of memorized material and got stuck.
Key Point
AI interviews do not just evaluate the "accuracy of the answer."
The way you prepare itself, in other words, how much you think in your own words, is what is being tested.
Instead of memorizing model answers, please construct your answers using your own episodes.
Difference ②: How to face the camera
What successful candidates were doing
Successful candidates adjusted the camera position to eye level.
Specific preparation:
When using a laptop, they placed it on a stand to bring the camera to eye level
When speaking, they looked at the camera lens rather than the screen (their own face or the interviewer's image)
To keep their gaze steady, they placed a small sticky note next to the camera that said "Look here"
This creates a state where it looks like "eye contact" from the interviewer's perspective.
What rejected candidates were doing
Many rejected candidates were looking at the screen.
Looking at the screen means looking slightly below the camera.
From the interviewer's perspective, it looks like "the gaze is off or looking down."
This is the same for in-person interviews, but the impact is particularly significant in online and AI interviews.
Key Point
Many AI interview systems quantitatively evaluate gaze stability.
If your eyes are not aligned with the camera, it is evaluated as "low gaze stability," which affects your score.
Before the actual interview, try recording a video of yourself to check.
Just checking whether your gaze is directed at the camera makes a big difference.
Difference 3: Speaking speed and rhythm
What successful candidates did
Successful candidates spoke at a pace of approximately 300 characters per minute.
This pace is:
a speed that is easy for voice recognition systems to pick up accurately
a speed that allows the listener to understand the content while listening
a speed where the connection of meaning is easy to understand when transcribed
It is also a good point to pause slightly after you finish speaking.
This ensures the AI system has time to process the audio.
What rejected candidates did
Many rejected candidates spoke at a faster pace due to nervousness.
When you speak too fast:
voice recognition does not function accurately (resulting in listening errors)
the content itself cannot be evaluated
pronunciation becomes unclear, and in some cases, the AI may judge it as "no response"
Conversely, if you speak too slowly due to nervousness, your answer might be cut off in time-limited questions.
Point
If voice recognition does not function accurately, the content itself cannot be evaluated.
No matter how wonderful the content you speak is, if the system cannot receive it accurately, it will not be reflected in your evaluation.
Be as conscious of your "speaking speed" as you are of the "content you speak."
Practice method:
Record your answers using your smartphone's voice memo app
Try converting them using a transcription app (such as Google Docs voice typing)
If it is transcribed correctly, your speed and rhythm are likely fine
Difference 4: Structure of the answer
What successful candidates did
Successful candidates were conscious of a structure that starts with the conclusion.
Answer format (STAR method):
Conclusion: "I can do XX" or "I have experience in XX"
Situation: What the situation was
Task: What was required
Action: How you acted
Result: What the outcome was
Speaking in this order makes it easier for the AI system to read the logical structure.
What rejected candidates did
Many rejected candidates spoke at length and left the conclusion for last.
"Um, first, the situation at that time was... from there I did this... and as a result, I was able to do XX"
The content is conveyed in this order as well, but it is likely to be rated lower in an AI interview.
Point
Many AI interview systems evaluate logical structure as a numerical value.
The structure of "Conclusion → Reason → Specific Example" is highly valued.
It is not about talking for a long time, but whether the structure is clear.
Guideline:
Keep your answer to each question within 90 to 120 seconds
State your conclusion in the first 10 to 15 seconds
Focus on only one episode
Difference 5: How to handle silence
What successful candidates did
When successful candidates needed time to think, they wouldsay, "Please let me organize my thoughts for a moment,"and then use 2 to 3 seconds of silence.
Having this phrase helps:
The AI system determines that you are "thinking"
It conveys that you are thinking even in videos that recruiters review later
It reduces the risk of the silence itself being judged as "no answer"
Many successful candidates prepared and practiced these "bridge phrases" in advance.
Examples:
"Please let me organize my thoughts for a moment."
"Please let me think for a moment."
"I have a few things in mind, so I will speak while I organize them."
What rejected candidates did
Many rejected candidates would get lost in thought andremain silent for 5 seconds or more.
In interactive AI interviews (such as SHaiN), long silences are sometimes processed as "no answer."
Once judged as "no answer," the score for that question drops significantly.
Key Point
Handling silence is especially important in AI interviews compared to in-person interviews.
Include a phrase indicating you are thinking, then start speaking within 2 to 3 seconds.
If you develop this habit during the practice stage, it will come out naturally during the actual interview.
A 1-week preparation schedule for AI interviews
I have organized 'what to do and when to do it.'
This is a schedule starting 7 days before the interview.
Day 1-2: Information gathering and organizing questions
Research the company's business, products, and news.
Articulate the differences from competitors.
List 10 frequently asked questions in AI interviews.
(Self-introduction, strengths, weaknesses, motivation for applying, what you want to do after joining, etc.)
Day 3-4: Verbalizing episodes
Create 5 personal episodes using the STAR method.
Prioritize episodes that fit the job and industry.
Use ChatGPT as a sounding board by asking, 'Please summarize this episode using the STAR method.'
Rewrite the resulting text into your own words.
Day 5: Mock interview
Conduct a mock interview with ChatGPT (you can use the prompt from the 37th article).
Actually speak your answers out loud (thinking in your head is not enough).
After answering, check 'what structure you used to speak.'
Day 6: Check via video
Record a video of yourself answering on your smartphone.
-
Points to check:
Is the speaking speed appropriate? (Not too fast or too slow)
Are you looking at the camera? (Not at the screen)
Is your expression too stiff?
Are there long silences?
Day 7: Final environment check
Adjust the camera position to eye level
Check the lighting (ensure your face isn't dark and you aren't backlit)
Check the microphone volume and sound quality
Check the stability of your internet connection
Check that the background of the room is not cluttered
Summary: What determines success or failure is not the 'amount of practice' but the 'correctness of practice'
Let's look back at the 5 differences.
① Preparation | Articulated for each company | Memorized model answers by rote
② Camera | Looked at the camera | Looked at the screen
③ Speaking speed | Spoke at a pace of 300 characters/minute | Spoke too fast due to nervousness
④ Answer structure | Spoke with the conclusion first | Put the conclusion at the end
⑤ Silence | Added a phrase like 'Let me organize my thoughts' | Continued silence for 5 seconds or more
Successful candidates have one thing in common: rather than just saying they "practiced a lot," they "practiced using the right methods."
Checklist (Confirm before the actual interview)
Articulating "Why this company?" for each specific firm
Preparing 5 personal episodes using the STAR method
Aligning the camera position with eye level
Checking speaking speed by recording a video
Practicing a conclusion-first response structure
Preparing "filler phrases" to use before silence
Finishing checks for audio, lighting, and background
Actions you can take right now
Today, record a 1-minute self-introduction video on your smartphone and watch it back
Check your speaking speed, camera angle, and the length of your pauses
Try doing one mock interview with ChatGPT using the prompt from the 37th installment
Please try starting even just one of these today.
If you found this article helpful, please give it a like.
Thank you for reading until the end.
A "like" takes only a second to press.
It helps reach other job seekers with the same concerns, so please do so.
I am also waiting for your comments.
"What was the most difficult or concerning thing you experienced when taking an AI interview?"
Please let me know. It will help with improving the article and choosing future topics.