There is no single pass mark. An AI interview is a structured assessment of what you say, so you pass by being specific: name the system, your part in it, the decision you made and the result. Answer in about a minute, expect a follow-up asking exactly what you did, and practise out loud before the real one.
What passing even means here
Nothing in an AI interview stamps you accepted or rejected. What comes out the other end is a written assessment, and what happens to it depends on the tool.
- In an employer's hiring process, your interview is turned into a report and a ranking, and a person reads the top of that list. HireVue, for example, tells candidates in its FAQ that its interviews and assessments "are intended to be first steps in the overall interviewing process" and that "those in the top group move on to person-to-person interviews before hiring decisions are made by people."
- On Career1, there is no pass mark at all. A finished vetting interview becomes your profile whatever the score, and companies searching the pool see it with the evidence attached.
So "passing" is not a threshold you clear. It is the difference between a transcript full of specifics that a reviewer can quote and a transcript of confident generalities they cannot.
What an AI interview actually evaluates
Almost all of the signal is in the content of your answers. That is not a reassurance, it is a warning: there is nowhere to hide a thin answer behind warmth or eye contact.
What is generally assessed. Whether you answered the question asked, whether the detail is real, whether your role in the story is clear, whether your reasoning holds up when pushed, and how clearly you explain something technical. Structured tools apply the same criteria to every candidate for a role, which is the main argument for them.
What is being assessed less than people assume. HireVue announced on 12 January 2021 that it had "made the decision to not use any visual analysis in our pre-hire algorithms going forward", saying visual analysis "has far less correlation to job performance than other elements of our algorithmic assessment". Some vendors never used video at all: Sapia.ai runs a structured chat interview, text only. Career1 records video for people to watch, and writes the report from the transcript, your resume and the job description; it does not analyse your face, expressions or appearance.
Career1 is not affiliated with HireVue, Sapia.ai or any other vendor mentioned here, and every claim above comes from those companies' own published pages. If you want to know what a specific employer's tool does, the invitation email and the vendor's candidate FAQ are the only reliable sources. Guides that tell you to smile more for the algorithm are guessing.
How to shape an answer that survives a transcript
Read your own answers back as text and the problem is obvious: most people say almost nothing concrete. A shape that fixes it, in about sixty to ninety seconds:
- One sentence of context. What the situation was and how big it was.
- What you did, in the first person, with the actual decision in it.
- The hard part, and why you chose one option over another.
- The result, with a number if you have one and an honest "I do not know" if you do not.
The same question, answered both ways:
- Weak: "I improved our reporting. It was pretty manual before, so I automated a lot of it and the team was much happier with the numbers."
- Strong: "Our support team rebuilt the same weekly report by hand every Monday, about three hours each time. I moved it into a scheduled query and a dashboard. The hard part was that two teams counted a reopened ticket differently, so I had to agree one definition with both leads before any of the numbers meant anything. It went from three hours to zero, and the definition argument stopped."
The second answer is not longer by much. It names the work, the obstacle, the decision and the outcome, and it gives an interviewer three specific things to ask about next. That is what you want.
Some habits do most of the work. Say "I" when you mean you and "we" only when you mean the team. Name tools, systems and sizes. Give the reason behind a choice, not just the choice. And never claim a skill you cannot discuss for two more minutes, because everything on your resume is a legitimate question.
The follow-up is the interview
A conversational AI interviewer chooses its next question from your last answer. It is told to push when something is vague, surprising or impressive, which means the question that decides your interview is usually the second one, not the first.
The follow-ups are predictable in kind, if not in wording: what exactly broke, how did you know it worked, what did you try first, what would you do differently, what did that cost you. Three ways to handle them well:
- Answer narrower, not longer. A follow-up is asking for one specific detail. Give that detail, then stop.
- Show the reasoning when you do not have the fact. "I do not remember the exact latency, but it was slow enough that support noticed before monitoring did" is a good answer. Inventing a number is not.
- Say when you have not done something, then reason about it. "I have not run a migration that size, but I would start by making the write path double-write" survives the next question. A bluff collapses on it.
Preparing for follow-ups is different from preparing answers. Pick five stories from your own work, then for each one make yourself answer "and what exactly did you do?" three times in a row. The third time is where the interview lives.
How people fail these interviews
Very few failures are technical. These are the recurring ones.
- Answering in headlines. Thirty seconds of summary with no detail reads as nothing on paper. A first answer under about forty seconds is usually too thin.
- Talking for four minutes. The opposite failure. You cover less, and a conversational interviewer with about eight minutes has to cut you off to get through its plan.
- Reading a script. It flattens your delivery, it does not answer the question that was actually asked, and it falls apart at the first follow-up. Notes with project names and numbers are fine. Sentences are not.
- Living in "we". A reviewer cannot tell what you did. This is the single most common reason a strong candidate scores as average.
- Pausing mid-thought. Spoken interviewers decide your turn has ended after a short silence, on Career1 well under a second. Think before you speak, and if the interviewer starts talking over your pause, just keep going: it is told to stop and listen when you interrupt.
- Trying to game it. Keyword stuffing an answer, or reading text generated during the interview, produces fluent answers with no specifics, which is exactly the pattern a follow-up exposes.
- Sorting out the microphone during question one. Fixable in five minutes beforehand, unfixable afterwards.
Set the room up, and know what to do if it breaks
Five minutes of preparation removes most of what goes wrong.
- A quiet room, a door that closes, and your phone on silent.
- Wired headphones with a microphone if you have them. Speakers echo your own voice back and confuse where your turn ends.
- The browser the invitation asks for, opened once in advance so you can grant microphone and camera permission before the interview, not during it.
- A wired or strong connection. HireVue's candidate FAQ, for example, recommends "an internet download and upload speed of at least 350 Kbps per second" for its on-demand interviews.
- Water, and your resume open on a second screen or on paper.
If the connection drops. Do not panic and do not start a new application. Recorded interviews are usually resumable: HireVue's FAQ says that "if you do lose connection in the middle of your assessment, you'll be able to pick up where you left off with the same link you used to begin the interview previously." Reopen the original invitation link first. If the tool will not let you back in, email the recruiter or the address on the invitation the same day, say what happened and when, and ask how to continue. Employers deal with this constantly and a same-day message is treated very differently from silence.
On Career1, a practice interview can simply be taken again the following week. A vetting interview is one attempt with no retakes, which is exactly why you should use practice first.
How to practise, and what practising cannot fix
The only preparation that moves the needle is answering out loud and then handling the second question. Start with the questions: the interview questions hub has the ones every interview shares and what a strong answer to each actually shows, and the AI interview questions guide covers what AI interviewers in particular tend to ask.
Then say your answers aloud, recorded, and listen back once. It is unpleasant, and you will hear the filler and the missing detail immediately.
Career1's free AI mock interview is a practice interview built from your own resume that speaks, listens and follows up. It is private: once a week, no video, no score, written feedback on what to work on, and never shared with any company. Practising with something that actually interrupts and asks "what did you do?" is the closest rehearsal to the real thing.
What practice cannot do is invent experience. If a role needs three years of something you have not touched, a better answer will not close that gap. Nor does it help if you are not reaching the interview: most filtering happens earlier, as does AI reject your resume? and remote jobs in the AI era explain.
Questions people ask
How do I pass an AI interview?
Answer the question asked, in about a minute, in the first person, with real detail: the system, your decision, the obstacle and the result. Expect a follow-up asking exactly what you did, and answer it narrowly rather than at length. Practise out loud first, because the difficulty is speaking, not knowing.
Can an AI interview tell if I use ChatGPT during it?
Career1 makes no claim to detect it, and neither should any guide. What happens in practice is simpler: generated answers are fluent and unspecific, so the follow-up questions about what you personally did have nothing to land on. On a recorded interview, a person can also watch you reading.
What happens if my internet drops during an AI interview?
Reopen the original invitation link first; recorded interviews are often resumable from where they stopped. If you cannot get back in, email the recruiter or the address on the invitation the same day explaining what happened. On Career1, practice can be retaken the following week.
Is it bad to say I do not know in an AI interview?
No, as long as you do not stop there. "I have not used that, but here is how I would approach it" gives a reviewer something to judge and survives the next question. A guess presented as fact is the answer that actually costs you.
How long should my answers be?
Roughly sixty to ninety seconds for a first answer, shorter for a follow-up. Under forty seconds usually means you have given a summary with no evidence in it; over two minutes and you are using time the interviewer needs for the rest of its questions.
Practise it out loud, in private
Career1's AI interviewer reads your resume, asks about your own work by voice and follows up when an answer is thin. A practice interview is private: once a week, no video, no score, and written feedback on what to work on. Career1 is free for job seekers.
The vetting interview is separate: one attempt, no retakes, recorded and scored, and it becomes a profile companies can find. You can hide that profile at any time.
Keep reading
- AI interview questions: what AI interviewers ask and how answers are scored What AI interviewers ask, how your answers are scored, and how to prepare, explained from the inside by the team that builds Career1's AI interviewer.
- How to prepare for a mock interview, out loud A practical routine for a mock interview: pick your stories, use STAR without sounding scripted, rehearse follow-ups, time answers and review yourself.
- Does AI reject your resume? What really happens to your application What really happens to a job application: how ATS parsing, AI screening and AI interviews differ, what rejects automatically, and what you can control.
- Remote jobs in the AI era: how remote hiring changed, and what to do about it How remote hiring works now software reads applications first, where the remote openings actually are in 2026, and what a strong application looks like.