Glossary
Conversational AI interview
A conversational AI interview is an asynchronous interview in which an AI asks spoken questions, listens to the answers and asks follow-up questions, much as a human interviewer would.
A conversational AI interview keeps the timing of an asynchronous interview, taken whenever the candidate likes, and adds the part one-way video leaves out: someone asking “what happened next?”
What the follow-ups add
Take this exchange from a support role interview:
Question: Tell me about a time you dealt with an escalation. Candidate: I handled a lot of escalations at my last job. I’m good at calming people down. Follow-up: Can you pick one and walk me through it? What was the customer upset about, and what did you do first?
The first answer is generic. The follow-up gives the candidate a second chance to show real experience, which is fairer to them and more useful to you. In a one-way video, that generic answer is all you’d get.
What it produces
Most tools produce a recording, a transcript and some kind of assessment. The useful ones rate each skill separately and show their reasoning, ideally with the words from the transcript that support each rating. A single overall number with no explanation is hard to trust and harder to defend.
Adaptive versus consistent
Follow-ups mean no two interviews are identical, which sits in tension with the idea of a structured interview. Two ways to handle it: keep the core questions fixed and let follow-ups only probe for detail, or switch follow-ups off for roles where strict comparability matters most. LessRounds offers both, as AI-led interviews with follow-ups and scripted interviews read word for word.
What to check before using one
- What is scored? The words of the answer, or also faces, voices and expressions? The latter adds little and raises real fairness concerns.
- Is there evidence? Can a reviewer see why a score is what it is?
- Who decides? The AI should prepare the shortlist. A person should decide who moves on.
- Are candidates told? They should know an AI is running the interview and that it’s recorded, before it starts.
- How does it handle language? Accents, grammar and mixing languages shouldn’t lower a score unless the job genuinely needs that skill.