One-way video interviews vs AI interviews.

One-way video interviews and AI interviews look alike at first. In both, a candidate opens a link, answers questions on camera or microphone, and your team reviews later. The difference is what happens between the questions. One records answers to fixed questions. The other listens and responds. That changes what you learn, how long review takes, and how the experience feels to the candidate.

How one-way video interviews work

You write a set of questions. Each appears on screen as text or a short recorded clip. The candidate gets some thinking time, then records an answer, usually with a time limit and a set number of tries. When they finish, your team watches the recordings.

Strengths:

• Every candidate gets exactly the same questions, which makes answers easy to compare.
• It’s simple. Nothing surprising happens during the interview.
• You see exactly what the candidate said, with no summary in between.

Weaknesses:

• Nobody asks a follow-up. If a candidate gives a vague answer, it stays vague.
• Talking to a camera with a countdown feels odd to many people, and some freeze.
• Review is slow. Someone still has to watch every recording, which uses up much of the time you hoped to save.

How AI interviews work

An AI interview is a spoken conversation. The AI asks a question, listens to the answer, and decides what to do next: ask a follow-up, move on, or bring the candidate back to the question if they wandered off it. Afterward it writes up the interview, usually with a score per skill and the evidence behind it.

Strengths:

• Follow-up questions turn short, vague answers into useful ones. “You mentioned you dealt with angry customers. Which one was the hardest?” is often where you learn the most.
• It feels closer to a real interview, which helps nervous candidates relax.
• Review is faster. You read a write-up and check the quotes, then watch the parts that matter.

Weaknesses:

• Follow-up questions vary between candidates, so no two interviews are exactly alike, unlike one-way video.
• You have to trust the write-up, so it needs to show the evidence behind each score.
• Some candidates are wary of talking to AI until you explain what it is and isn’t doing.

Side by side

Some tools offer both. An AI interview can run in a scripted mode that reads every question word for word. Everyone gets the same questions, as with one-way video, but it still feels like a conversation and you still get a written summary.

When one-way video fits

• Presentation is the skill. In some roles, how someone comes across on camera matters, and your team wants to judge that directly.
• You need identical interviews, for example when several teams will compare candidates across one large group of applicants, and follow-up questions would make that harder.
• Your volume is low enough that watching every recording is fine.
• You want candidates to prepare. Fixed questions with thinking time suit short, prepared answers, like a sales pitch.

When an AI interview fits

• You have more candidates than time to watch recordings. Reading a write-up is quicker than watching every minute.
• The skill shows in how someone thinks, not how they present. Troubleshooting, prioritizing, handling customers and explaining processes all benefit from a follow-up question.
• Candidates are early in their careers and give short first answers. A good follow-up draws out what they actually know.
• You hire in more than one language and want each candidate to answer in the language they’re most comfortable with.
• You want every score explained, so reviewers can check the evidence instead of rewatching everything.

Questions to ask before you choose

1. What do you want to learn in the first interview? If it’s “can they explain their thinking”, you want follow-up questions.
2. Who will review, and how much time do they have? Multiply the number of finished interviews you expect by the review time for each one.
3. Does the skill need a camera? If not, audio-only interviews are often more comfortable for candidates and just as useful.
4. How will you explain it to candidates? Either format needs a clear invite. Our interview invite template is a starting point.
5. What does the write-up show? A number with no reasoning is worse than a recording.

Making either format work for candidates

The format matters less than how you set it up. A few habits help with both:

• Say how long it takes, honestly. “About 15 minutes” in the invite means candidates start when they have 15 minutes, not in a two-minute gap between other things.
• Say who, or what, they’ll be talking to. Candidates who expect a person and get an AI, or expect a conversation and get a countdown timer, start the interview flustered.
• Keep questions short. If a question takes 40 seconds to read or hear, it’s probably two questions. Split it.
• Allow a phone. Many candidates, especially for hourly and early-career roles, don’t have a laptop handy. Test your interview on a phone before you send it.
• Tell them what happens next. “We review every interview within two working days and will email you either way” takes away most of the worry about what happens after they finish.

If you’re unsure which format candidates will prefer, run a small test: send each format to a small group for the same role and ask a few people from each how it felt. Their answers are usually clearer than any feature list. Our guide on candidate experience in async interviews covers this in more depth.

LessRounds runs AI interviews in either mode: AI-led, with follow-ups, or scripted, reading your questions word for word. Both can be video or audio-only, and every skill score comes with its reasoning and a quote. See the AI interviewer for how a conversation runs.