A client called me last week, a little rattled. She'd just finished a screening interview for a role she really wanted. The interviewer was warm, asked smart follow-up questions, even laughed at the right moments.
It wasn't a person.
She found out afterward — buried in a follow-up email — that she'd spent twenty minutes talking to an AI. And she's far from alone. According to Greenhouse's 2026 candidate experience report, 63% of job seekers have now faced at least one AI-conducted interview, and 70% weren't told upfront that the "interviewer" wasn't human.
If that makes your skin crawl a little, I understand. But here's what I told my client, and what I'll tell you: this is the new front door. You can resent it, or you can learn to walk through it better than everyone else. Let's do the second one.
What an AI interview actually looks like
"AI interview" covers a few different experiences, and it helps to know which one you're facing:
- One-way video interviews. You record answers to preset questions on your own time. Software analyzes what you say — and increasingly, how you structure it.
- AI voice or chat screeners. A conversational bot calls or messages you, asks screening questions, and adapts its follow-ups based on your answers. These are the ones people mistake for humans.
- AI-assisted live interviews. A real recruiter runs the conversation, but AI transcribes it and scores your answers against the role afterward.
In every version, the same thing is true: your answers are being converted into data and compared against the job description. Which means the way you prepare has to change.
Why companies are doing this
It's not laziness — it's volume. When a single opening draws hundreds of applications (and in this job market, it does), companies use AI to screen more candidates than any recruiting team could reach by phone. A screener that talks to every applicant, instead of the first thirty, is genuinely fairer in one sense — everyone gets a conversation.
The catch is that the conversation is being graded by software. And software grades differently than people do.
The AI is listening for what the humans skim for
Here's the mental shift that helps my clients most: an AI interviewer is doing to your spoken answers what an applicant tracking system does to your resume. It's checking your language against the language of the job posting — the skills, the tools, the responsibilities.
If you've read my post on how I make sure resumes pass the ATS, this will sound familiar: the postings themselves tell you what the machine is scoring. Before any AI interview, pull up the job description and highlight the terms that appear more than once. Those words belong in your answers — naturally, not robotically.
How to ace it
Treat it exactly like a human interview. Dress the part, sit somewhere quiet and well-lit, and bring real energy. It feels strange to be warm and animated for software, but your tone, pacing, and confidence still register — and if a human reviews the recording later (they often do), flat delivery reads as low interest.
Structure every answer. AI scoring rewards organized responses. Use the STAR frame: the Situation, the Task, the Action you took, and the Result. Rambling hurts you more with an algorithm than it ever did with a person, because the software can't read between the lines the way a generous interviewer would.
Say the actual words. If the posting says "stakeholder management" and you say "keeping everybody in the loop," a human knows those match. Software may not. Use the posting's own vocabulary for your skills — then prove it with a specific example.
Answer the question that was asked. AI follow-ups are generated from your responses. Vague answers produce vague follow-ups, and the interview drifts. Specific answers keep the conversation on your strongest ground.
Don't try to game it. Candidates trade "hacks" online — hidden keywords, scripted monotone answers optimized for transcription. Skip all of it. The systems get better every quarter, and anything that feels gamed to the software will feel worse to the human who reviews the finalists.
Test your tech, then forget it. Bad audio has sunk more AI interviews than bad answers. Use headphones with a decent microphone, close every other tab, and do a thirty-second test recording first.
It's okay that this feels weird
That same Greenhouse report found that 38% of candidates have walked away from a hiring process because of AI interviews. I don't blame a single one of them — being screened by software for a job you'd pour yourself into can feel dehumanizing, especially when nobody told you it was coming.
You're allowed to ask questions. "Will this interview be conducted or evaluated by AI?" is a fair and professional thing to ask a recruiter. So is "Will I have the chance to speak with a person before a decision is made?" Good companies answer plainly. Evasive answers are data, too — about how they treat people generally.
But I'd gently push back on walking away on principle alone. The AI screen is usually the first gate, not the whole castle. The people you'd actually work for are still on the other side of it, and they never meet the candidates who opt out in round one.
The machine screens. The human hires.
Every AI interview eventually leads to a person — a hiring manager who wants to know what the software can't measure: whether they'd trust you, enjoy working with you, bet on you. So prepare for both layers. Clear, structured, keyword-aware answers to get through the screen. Real stories, warmth, and presence for the humans who make the decision.
That two-layer preparation is exactly what I work on with my interview coaching clients — we rehearse the structured answers the software rewards and the storytelling that wins over the room. If you've got an AI interview on your calendar and you're not sure you're ready, book a free discovery call — twenty minutes by phone, and you'll walk away with a plan.
The interviewer might not be human. But the person they end up hiring will be. Make sure it's you.