# Only 7% Would Refuse an AI Interviewer. Rehearse Your Resume Out Loud

_AI voice screens like HireVue, Ribbon and Tenzo score knockouts pass/fail and skills 1 to 5. Here is how to rehearse so your answers match your resume._

- Canonical URL: https://www.refolk.ai/candidates/blog/ai-voice-screen-interview-resume-rehearsal
- Published: 2026-11-02
- Reading time: 10 min
- Keywords: how to pass an AI interview, AI voice screening interview tips, HireVue AI interviewer questions, AI phone screen what to expect, AI recruiter interview preparation

If HireVue, Ribbon or Tenzo just invited you to a voice screen, your voice matters less than you'd expect. What the screen really checks is your resume, read back to you out loud, and your score depends on whether your answers match what you wrote.

You will probably take the screen anyway. A Greenhouse report published Oct 21, 2026 surveyed more than 2,800 job seekers across the U.S., U.K., Ireland, Germany and Australia. Only 7% would refuse an AI-led interview outright, and candidates were 2.7x more likely to opt for an AI interview than to withdraw if a human interviewer wasn't available.

One caveat on the source: Greenhouse sells hiring software, including a voice AI interview product, so it has a stake in candidates accepting the format. The numbers are still the best recent read on how candidates feel about it. Nearly half (48%) of respondents had taken an AI interview in the past year, and 70% of those said it let them demonstrate their skills well.

## What does an AI voice screen actually score?

An AI voice screen scores two kinds of answers: qualification questions, marked pass or fail, and skill questions, scored on a rubric that usually runs 1 to 5. Both are scored before a recruiter opens your file.

An **AI voice screen** is an automated first-round interview. Software asks spoken questions, transcribes your answers and scores them against a rubric the employer has approved. The main tools set it up differently:

| Tool | Where the questions come from | How answers are scored |
|---|---|---|
| HireVue AI Interviewer | AI drafts from the job description, employer approves | Qualifications: Meets / Does not meet. Skills: 1 to 5 rubric |
| Ribbon | The job posting (Tesla candidates describe an audio-only screen) | Communication, enthusiasm and motivation, skills and experience, each out of 5 |
| Tenzo | Your resume plus the role | Knockouts on authorization, certifications, availability, location; answers evaluated alongside the resume |

HireVue ties its skill scores to **behaviorally anchored rating scales**, which are rubrics where each score level describes what an answer at that level sounds like. That structure rewards answers that name specific actions and results. Fluent but general answers don't climb it.

Ribbon's own sample interview starts with a work-eligibility question and then asks about your start date. Those are both facts you already put on paper.

## Why is the AI screen really a resume audit?

Resume-aware screeners use your resume as their question list. Each bullet, number and date becomes a prompt, and the score depends on whether your spoken answer supports it.

Tenzo says this outright. In its comparison against HeyMilo, Tenzo says it "dynamically tailors its questions based on each candidate's resume," which means surfacing red flags, confirming key experiences and digging into gaps. Warden AI, its bias auditor, describes the tool as evaluating candidate responses "along with their resume to assess job fit." A third-party guide to Tenzo's AI recruiter, "Morgan," tells candidates to keep their resume fresh in mind because Morgan references it.

The reason this matters: once the system holds both the document and the transcript, any contradiction between them is a signal it can detect. Suppose your resume says you "cut the support backlog by a third" and you tell the screener "we brought it down a fair bit." That isn't a stylistic slip. Your answer just scored lower than the claim it was meant to back up.

Tools that write questions from the job description, like HireVue, still lead you back to your resume. When a skill question asks for an example, you pick it from the work you listed. If the example isn't on the page, or doesn't match it, the mismatch is there in the record.

## Who reads your result after the screen?

The recruiter who reads your result probably never heard the interview. They see a summary and a tier label. HireVue itself recommends that employers show result tiers (top, middle, bottom) and keep numeric competency scores switched off.

Refolk's index of professional profiles shows how far the tool sits from the recruiters who rely on its output:

| Segment | Count | Source |
|---|---|---|
| US recruiters, TA staff and sourcers listing HireVue as a skill | 16 | Refolk's index |
| UK recruiters, TA staff and sourcers listing HireVue as a skill | 0 | Refolk's index |
| US recruiters, TA staff and sourcers with "AI recruiting" in their headline | 45 | Refolk's index |
| HireVue-skill profiles per "AI recruiting" headline profile, US | 0.36 | Derived (16/45), separate queries |
| Job seekers who would refuse an AI interview | 7% | Greenhouse, Oct 2026 |

> **16** US recruiters, TA staff and sourcers in Refolk's index who list HireVue as a skill
>
> Their employers are high-volume, non-coding hirers such as Novant Health, Takeda and Securitas Security Services USA.

These counts are small. They show that recruiters rarely list the tool as something they know. They don't tell you how widely it's used. Ameritas and Quantum Health show up on the same short list, which is the pattern you'd expect: high-volume, non-coding hirers are where a screener earns its cost.

For you, the takeaway is that the screening tool is plumbing. The recruiter receives what comes out the end: a label and a few lines of summary. A mismatch between your answer and your resume becomes one sentence in that summary ("candidate unclear on certification status"), and you won't be there to explain it.

## Why do knockout questions punish hedging?

Knockout questions are scored pass or fail, so a hedged answer earns no partial credit. "Meets / Does not meet" has nothing in between.

A **knockout question** is a yes-or-no eligibility check that can remove you from the pipeline on its own, whatever your other scores are. Tenzo's knockouts cover work authorization, certifications, availability and location. Say your resume reads "certified" and you tell the screener "I'm working toward it." The rubric now has two conflicting facts, and only one of them meets the bar.

Before the screen, list every hard claim on the resume you submitted and write a one-sentence answer for each:

- **Work authorization:** your exact status, and whether you need sponsorship now or later.
- **Certifications and licenses:** the exact name, whether it's active, and the year earned or expected.
- **Availability:** a real start date or notice period, not "flexible."
- **Location:** where you live now, whether you'll relocate, and which days on-site you can commit to.
- **Employment dates:** month and year for each role, matching the resume exactly.
- **Titles:** the title on paper, plus one clause on scope if it undersells the job.
- **Years of experience:** the number the posting asks about, counted the same way your resume implies.

Run this list before you apply, too. [Refolk](/candidates) scores how well you actually fit a posting, so a hard requirement you don't meet shows up while you can still decide whether to apply, not when a screener asks about it.

## Do tone and delivery matter in an AI voice screen?

Delivery matters less than most candidates assume, and the facts matter more. HireVue's Explainability Statement says its AI "relies only on what is said by the candidate" and does not assess facial expressions, body language, background and surroundings, or tone of voice.

That changes how you should spend your prep time. An hour practicing a confident delivery in the mirror buys you less than an hour rehearsing the numbers behind each bullet. Ribbon does score communication and enthusiasm by default, so delivery isn't worth zero. But the skills-and-experience score is the one tied to your resume, and that's where vague answers do the most harm.

Don't swing the other way and read from a script. Answers that sound recited tend to get flagged by scorecard tools. The goal is to know your facts well enough to say them in plain sentences. Memorized paragraphs won't get you there.

For AI voice screening interview tips that actually move the score, rehearse each resume bullet in this order:

1. What the situation was, in one sentence.
2. What you personally did, with a verb and a tool or process name.
3. The number, its baseline, and how it was measured.
4. What changed afterward.

## What does the Graham Dot Zip parody get right?

The viral mid-October video of an AI agent interviewing with an AI recruiter was staged, but the failure it shows is real: the candidate's AI representative overstated his experience. That is exactly how people fail a resume-aware screen.

The creator, Graham Dot Zip, called the video a parody. He told Lead Stories, whose Alan Duke fact-checked it, that the videos are "staged, but not scripted." The reported post date varies (Oct 12 or Oct 13 depending on the outlet). Treat it as satire, not footage of a real interview. In it, Graham's AI "representative," Victor, inflates his work history even as Graham tries to correct him.

Take the agent out and the lesson still applies. A screen built to confirm key experiences and probe gaps will find the gap between what you claimed and what you can explain. Ophir Samson, Greenhouse's head of voice AI, puts it this way: "AI can polish a resume or fire off hundreds of applications, but it can't show an employer how someone thinks or learns." Again, his company sells this product. The logic holds anyway.

> An inflated resume you can't defend out loud costs you more than a modest one you can.

This is why it matters where your resume came from. Refolk writes your resume from your own history and tailors it to each posting without inventing experience, so every line it produces is one you can explain when Morgan or HireVue asks about it.

## How should you prepare for an AI recruiter interview?

The best AI recruiter interview preparation is an out-loud rehearsal of the exact resume version you submitted, with the job posting open next to it. Here's how to pass an AI interview without contradicting yourself:

1. **Pull up the exact version you sent.** If you tailored it, the screener is reading that copy, not your master resume.
2. **Answer the knockouts first.** Use the checklist above. Say each answer out loud once and cut any hedging words ("kind of," "I think," "almost").
3. **Map requirements to stories.** For each skill in the posting, pick one resume bullet that proves it. Skill questions score 1 to 5, and a specific story is what moves you up the scale.
4. **Defend every number.** For each figure on the page, know the baseline, the timeframe and your share of the credit.
5. **Prepare your gap answer.** Tenzo says it dives deeper on gaps. Give dates to the month, one sentence on the reason, and one on what you did during that time.
6. **Record and replay.** Listen for answers that stray from what the resume says. Fix the answer or fix the resume, but don't let them disagree.
7. **Set up for audio.** Ribbon's Tesla screen is audio-only, as Tesla candidates describe it. A quiet room and a decent headset keep the transcript clean, and the transcript is what gets scored.

As for what to expect from an AI phone screen in practice, Ribbon's sample is a fair template: eligibility first, then a start date, then skill questions that ask you to back up a claim.

## Does passing the AI screen mean you'll hear back?

No. Completing an AI interview gets you through a filter. It doesn't guarantee anyone will talk to you. In Greenhouse's May 2026 data, among people who completed an AI interview, 28% moved forward, 13% were formally rejected, and the rest mostly heard nothing.

> **51%** Candidates who completed an AI interview and never heard back
>
> Greenhouse, May 2026. Only 28% moved forward to the next round.

Tenzo claims candidates who finish its pre-screen are 3 to 5 times more likely to hear back, but there's no guaranteed feedback or human response. Greenhouse's October report adds that 81% of US candidates were rejected from a job in the past year without speaking to anyone at the company, human or AI.

So a well-prepared screen can still disappear into silence. You rehearse every date, defend every number, clear the knockouts, and then wait for nothing. Do that across a dozen applications and the voice screen becomes one more unanswered step in a process that rarely gets back to you.

None of that makes the prep a waste. A clean, consistent screen gives you the best shot at the 28% rather than the 13%. Just don't take the screen's friendly tone as a sign that a person is listening yet.

## FAQ

### What HireVue AI interviewer questions should I expect?

Expect two types. Qualification questions check eligibility, such as work authorization, required certifications or location, and are scored Meets or Does not meet. Skill questions ask for examples of the competencies in the job description and are scored on a 1 to 5 rubric. HireVue's AI drafts both from the job description and the employer approves them, so the posting is your best preview of the questions.

### Can I let an AI assistant answer the screen for me?

The Graham Dot Zip parody shows why that's a bad idea. Its AI representative overstated his experience even while he tried to correct it. Resume-aware screeners like Tenzo are built to confirm experiences and probe gaps, and answers that sound recited tend to get flagged. Prepare your own facts and say them yourself.

### What if my resume claims something I can't back up out loud?

Fix the resume before the screen if you still can, because knockout questions are pass or fail and a hedged answer can fail outright. If the version you submitted is already with the employer, answer truthfully and precisely ("I passed the exam in March, the license issues in June"). A precise correction beats a vague answer that contradicts the page.

---

_From the Refolk job search blog. More posts: https://www.refolk.ai/candidates/blog/index.md_