Is your interview scored by AI? How to tell, and what it means

A one-way video interview is sometimes scored by AI, and often it is not. Plenty of the interviews people call “AI interviews” are simply recorded answers that a person watches later. When AI is involved, it usually transcribes what you say and scores the words against the employer’s criteria. A few tools have also analyzed face and tone, though that practice has faded and several places now restrict it by law. What gets measured depends on the tool the employer picked, and none of it is as mysterious as it can feel from the recording screen.

The unease is understandable. You sit alone, record into a webcam, and hear nothing back. It is easy to picture an algorithm grading your every blink. But the everyday reality is more ordinary than that, and the ordinary version works in your favor: in most cases the thing being judged is whether you actually answered the question. This guide walks through what AI scoring does, how to tell which kind of interview you are in, and the questions you are entitled to ask before you record.

What AI scoring actually does

In a typical AI-assisted one-way interview, here is the chain of events after you submit:

1. The tool transcribes your spoken answer into text.
2. It scores that text against the criteria the employer wrote, things like whether you gave a specific example, named a real result, or answered the actual question.
3. It produces a summary, a match score, or a short highlight clip for a person to review.

The thing being judged is mostly your words. Did you answer the question. Was there a real example in there. Did you make your point or ramble. This is closer to a graded transcript than a lie detector. If you give a clear, specific answer, the transcript carries it regardless of how nervous you looked.

It is worth knowing why employers use this step at all, because it tells you what they are looking for. Scoring the transcript against a fixed set of criteria means every candidate answers the same questions and gets read against the same yardstick, which is more consistent than a rushed live call where the interviewer’s mood and memory do a lot of the work. It also lets a small team get through a large applicant pool without quietly dropping people for lack of time. Used well, the AI is a first-pass reader that helps a human spend their attention on the answers worth a closer look. You are usually being organized and summarized, not sentenced.

What it can read, and what it cannot

Some tools, especially older ones, layered video and audio analysis on top of the transcript. One recruiter described HireVue using AI to analyze recorded answers “including things like eye contact, body language” to score candidates, and openly wondered about bias, “especially as it relates to neurodiverse and international candidates.” That is a fair concern, and it is widely shared. It is also a big part of why this kind of analysis has been pulling back.

What these systems cannot do is read your mind, detect lying, or measure your worth. The claim that a webcam can score “employability” or “emotional intelligence” from your face drew sustained criticism from researchers and regulators, and the industry has largely moved away from it. One recruiter who used HireVue around 2020 put it plainly: the facial-recognition AI was polarizing, and “most if not all don’t use that module.” HireVue has also said publicly that it removed facial analysis from its scoring. So treat face-scanning as the exception now, not the default, and treat the transcript as the main event.

How to tell which kind you are in

You usually cannot tell from the recording screen alone, but here are the honest signals:

• The invite or consent screen says so. Many tools now disclose AI use before you record, partly because the law increasingly requires it. Read that screen instead of clicking past it. It often answers the whole question.
• The brand markets AI scoring. Some platforms, like HireVue, and the newer tools that present an animated or voice “interviewer,” build automated scoring into the product. A plainer record-and-submit format, the kind used by tools such as Spark Hire or Willo, is more often there to capture video for a person to watch. Check the tool’s own site if you want to be sure what it does.
• Whether a human appears in the early steps. A fully automated voice that asks questions and reacts is an AI interview. A static list of text questions you record against is usually a recording that a person reviews later.

When you genuinely cannot tell, assume your words are doing the work and answer accordingly. That assumption is right far more often than it is wrong, and it points you at the thing that actually moves the result.

Your consent rights

You have more standing here than most people realize, and it is growing.

Illinois has had an Artificial Intelligence Video Interview Act in effect since 2020. It requires an employer who uses AI to analyze a video interview to do three things before you record: tell you AI is being used, explain in plain terms how it works and what it evaluates, and get your consent. The law also limits how long recordings can be kept and shared.

Other places are moving the same direction. New York City requires bias audits of automated hiring tools under Local Law 144. Colorado has passed a broader AI-in-hiring law that phases in over the next couple of years. You do not need to memorize the statutes. The point is that “is AI scoring me, and how” is a fair question, and in a growing number of places employers are legally obligated to answer it.

So ask if you want to know. A short, neutral email to the recruiter is completely reasonable:

“Quick question before I record. Does the interview tool use AI to evaluate my answers, and if so, does a person also review them? Happy to proceed either way, I just like to understand the process.”

A good employer answers without flinching, and most do. The answer tells you a little about how they hire, and either way you go in informed.

How to handle it well

For most roles, an AI-assisted screen is a reasonable first step, and the way to do well in it is refreshingly low-tech. Stop picturing a machine grading your face and treat the recording like what it is: a first round where a clear, specific answer wins. Look at the lens. Answer the question that was asked. Give one real example with a real result. The transcript will carry you, whether a person reads it, an algorithm summarizes it first, or both.

A few candidates decline these on principle, and that is a legitimate personal choice. One person on Reddit wrote, after getting an invite, “I noped out.” That is one reaction, and online forums tend to collect the strongest ones. It is not the common experience, and it is not a verdict on the format. Far more people simply record their answers, get screened, and move on to a live conversation in the next round. If a one-way screen genuinely conflicts with what you want from an employer, walking away is a real option. For most situations, though, treating it as a normal early step is the move that keeps you in the running.

If the nerves are the real problem here, more than the AI itself, read how to handle one-way video interview anxiety. And if you want to know whether a person actually watches these, see do employers actually watch one-way interviews.