AI interviews have gone mainstream fast. Roughly two-thirds of organizations now use AI somewhere in recruitment, driven by sheer volume: job seekers submit close to 11,000 applications per minute on LinkedIn, up 45% in a year, and 70% of hiring teams say fewer than half of applications even meet the role's criteria. When a single opening draws hundreds of applicants, interviewing everyone fairly is impossible by hand. That is the problem AI interviews exist to solve.

‍

- What an AI interview actually feels like
- The two types of AI interview (and why the difference matters)
- AI interviews vs traditional interviews
- Are AI interviews biased?
- What real candidates say
- How to prepare for an AI interview
- Frequently asked questions

‍

Most people picture something cold and robotic. A well-designed structured AI interview is closer to a focused, unhurried conversation.

Here is roughly how a Hubert interview goes for a candidate:

‍

1. You get a link, and you choose when. No scheduling back-and-forth. You start the interview whenever and wherever suits you, on your phone or laptop. The median time to start is about a minute.
2. You have a conversation, not a form. The AI, Hubert, asks role-relevant questions one at a time and asks natural follow-ups based on your answers, so it feels like being listened to rather than filling in boxes.
3. You answer at your own pace. There is no camera studying your face and no awkward silence being timed. You can think before you respond.
4. Every candidate gets the same interview. The questions and the way answers are scored are consistent for everyone applying, which is what makes the process fair.
5. A recruiter makes the final decision. Hubert produces a scored, explainable shortlist; a human decides who moves forward.

‍

Here is a simplified example of a single exchange, to make it concrete:

‍

Hubert: 'This role involves handling several customers at once during busy periods. Tell me about a time you had to juggle competing priorities. What did you do?'

‍
Candidate: During the holiday rush at my last job, three tables and a phone order all needed attention at once. I acknowledged each customer so nobody felt ignored, handled the time-sensitive food order first, then worked through the rest in order of urgency.

‍
What Hubert assesses: the answer is scored against the defined competency (prioritization under pressure), using the same criteria applied to every other candidate for this role. The recruiter sees the score and the reasoning behind it, tied to this specific answer.

‍

Want to see the recruiter side in action? See a live demo

‍

Not all AI interviews work the same way, and the distinction is the single most important thing to understand about the category.

‍

AI-assessed video interviews. The candidate records video answers to preset questions, and software scores them, historically on facial expressions and tone as well as content. Two problems have dogged this approach. First, many of these systems are trained on a company's existing employees, so the model learns to favor candidates who resemble past hires, reproducing whatever bias is already in the workforce. Second, they often ftypically return a score with no explanation, a "black box" a candidate cannot understand and a recruiter cannot audit. These systems have drawn regulatory complaints and criticism precisely because they are hard to see inside.

‍

Structured, conversational AI interviews. Instead of scanning a face, the AI conducts a consistent, competency-based conversation and assesses what the candidate actually says against defined criteria for the role. Hubert takes this approach and scores answers with deterministic AI models: the same candidate giving the same answers gets the same score every time, and every score ties back to a specific response. That consistency is what makes the result explainable and auditable, rather than an opaque number. There is no facial analysis, and the final decision always stays with the recruiter.

‍

The lesson from the category's most famous failure still applies: when Amazon scrapped its AI recruiting tool, it was because the model had learned the company's historical bias and no one could see it happening. An AI interview is only as fair as its design allows you to verify.

‍

The clearest way to understand a structured AI interview is to hold it up against the traditional phone or screening interview it replaces at the top of the funnel.

The first difference is consistency. A traditional interview varies with the interviewer, their mood, and the time of day, so two equally strong candidates can have very different experiences. A structured AI interview gives every candidate the same questions, scored the same way. Availability differs too: a phone screen is limited to working hours and calendars, while an AI interview runs around the clock, so candidates start whenever it suits them. That feeds directly into scale, because a recruiter can only call so many people in a week, whereas an AI interview can screen hundreds or thousands without adding headcount.

The differences that matter most, though, are about fairness and accountability. A traditional interview is exposed to unconscious bias, from accent to appearance to how much rapport the interviewer happens to feel, whereas a structured AI interview assesses each candidate on their answers to defined competencies. It is also gentler on the candidate: instead of performing live under pressure, they answer at their own pace with no camera judging them. And where a human interviewer's reasoning is rarely written down, every AI interview score ties back to a specific answer, which means the decision can actually be explained and audited. That carries through to feedback, which is too often absent or generic after a phone screen but is structured, consistent, and defensible with a well-designed AI interview.

The point is that AI should not replace the human interview. It is that the initial screen, the stage that is slowest and most exposed to bias, is exactly where structure and consistency help most, freeing recruiters for the later conversations that genuinely need a human.

‍

This is the objection that matters most: an AI interview can be biased, and some are. If a system is trained on biased historical hiring data, it will reproduce that bias. Independent research keeps confirming it. In 2024, University of Washington researchers found the models behind many resume screeners favored white-associated names in 85.1% of cases and disadvantaged Black male candidates in 100% of cases tested. In October 2025, Stanford researchers found AI screeners rated older and female candidates lower than younger male candidates with otherwise identical resumes.

So the honest goal is not "zero bias". It is bias that is actively tested for, measured, and reducible, with every decision traceable. Two design choices make that possible:

‍

- Skills-based assessment instead of history-based matching. Scoring answers against the competencies a role requires, rather than looking for people who resemble past hires, avoids the trap Amazon fell into.

‍
- Deterministic, explainable scoring. Because Hubert produces the same score for the same answers every time and ties each score to a specific response, the result can actually be audited. A system that returns different results on different runs cannot.

‍

Under the EU AI Act, AI used in recruitment is classified as high-risk, and deployers face obligations including bias testing, human oversight, transparency to candidates, and record-keeping, with penalties up to EUR 15 million or 3% of global annual turnover. Fairness that you can demonstrate is becoming mandatory.

Weighing AI interviews for your team? Talk to us about implementation

‍

The most common worry recruiters raise is not accuracy or compliance. It is whether candidates will accept an AI interview at all.

The evidence from Hubert's own candidate reviews is reassuring: Hubert's average Google rating sits at 4.8 stars, and the same themes come up again and again, clarity, fairness, and low pressure.

‍

"I felt I was in the hands of a fair interviewer. There was no bias, it was practical, clear language and clear questions." - Candiate, UK)

"I really got the chance to think through my answers without feeling stressed." (Candidate, Sweden, translated from Swedish)

‍

One word appears across Swedish, Spanish, French, German, and Portuguese reviews alike: simple, "Smidig och snabb" (smooth and fast). "Muy buena experiencia" (very good experience). "Très bon process et simple" (very good and simple process). Candidates in different countries and languages independently reaching for the same word says something about the experience.

‍

That perception is not a happy accident; it reflects the Hubert Candidate Pledge, a C-level-signed commitment that governs how every interview is built. Across deployments, Hubert sees candidate satisfaction scores of 9/10 and completion rates above 85%. Candidates who find the process fast and fair finish it, which means more of the pipeline you built actually reaches the shortlist.

‍

If you are a candidate about to take one, the good news is that the best preparation is the same as for a good human interview: answer honestly and think about your actual experience.

- Treat it like a real conversation.

‍Answer as you would a person. With a structured interview, there is no "trick" to game; clear, specific answers score best.

‍
- Use concrete examples.

‍When asked about a situation, describe what you did and what happened. Specifics beat generalities.

‍
- Find a quiet moment.

‍You choose when to start, so pick a time you can focus.

‍
- Don't stress about how you look or sound.

‍A structured AI interview assesses your answers, not your appearance or accent.

‍
- Take your time.

‍You can think before you respond. Use that.

‍

An AI interview is only as fair and useful as its design lets you verify. The kind worth adopting is structured, skills-based, explainable, and built to keep a human in the decision, screening at scale without trading away fairness or defensibility. That is what Hubert is built to do.

See how structured AI interviews work for your team: book a live demo

‍

‍

What is an AI interview?
An AI interview is a job interview conducted or assessed by artificial intelligence rather than, or before, a human recruiter. The two main types are AI-assessed video interviews, where software scores recorded answers, and structured AI interviews, where an AI conducts a consistent, competency-based conversation and scores each answer against defined criteria.

How do AI interviews work?
It depends on the system. Structured AI interviews like Hubert's ask every candidate the same role-relevant questions, then score the answers against defined competencies using deterministic models, so the same answers always produce the same score, tied to a specific response. A recruiter then reviews the scored shortlist and makes the final decision.

Are AI interviews biased?
They can be, if trained on biased historical data. The way to manage bias is skills-based assessment, explainable scoring that can be audited, and continuous bias testing, rather than a vendor claim to have removed bias entirely. From August 2026 the EU AI Act requires bias testing and human oversight for hiring AI.

Do AI interviews use facial recognition?
Some AI-assessed video interview tools analyze facial expressions and tone, an approach criticized for opacity and bias. Structured AI interviews like Hubert's do not; they assess the content of a candidate's answers, not their face.

Do AI interviews replace human recruiters?
No. The effective model is augmentation: AI handles consistent, high-volume screening and produces a shortlist, while a human recruiter makes the final hiring decision.

How do I prepare for an AI interview?
Treat it like a real conversation, use concrete examples from your experience, choose a quiet time to start, and take your time answering. A structured AI interview assesses your answers, not your appearance.

‍