An AI Video Interview May Not Show The Real Candidate

I believe in AI.

I also believe an AI video interview may not show the real candidate.

This is the point that many employers are missing.

A candidate can be excellent at the job and still be weak in a recorded video. Another candidate can be average at the work and still look polished, confident, and perfectly prepared for the format.

That does not mean video has no use. It means video should not be treated as the truth.

When a company makes an AI video interview a mandatory first filter, it may not be measuring the candidate's actual ability. It may be measuring camera confidence, home privacy, internet quality, language comfort, lighting, anxiety, accent, and the ability to perform in front of a machine.

Those are not the same thing as job competence.

I Am Not Against AI In Recruitment

This is not an anti-AI position.

AI can be useful in recruitment. It can organize information, summarize CVs, help build structured rubrics, compare evidence against job requirements, support note-taking, and reduce some of the repetitive work that slows hiring teams down.

Used correctly, AI can help recruiters become more consistent and more prepared.

But AI should assist the hiring process.

It should not become the interview itself.

It should not become the hidden judge.

And it should not become the only gate a candidate must pass before a human takes responsibility.

The Interview Is Not Only The Answer

A real interview is not only a set of answers.

It is a conversation. It has context. It has follow-up questions. It allows a person to clarify an idea, explain a project, correct a misunderstanding, or show how they think under a normal human exchange.

A recorded video removes much of that.

The candidate is speaking to a screen. There is no human feedback. There is no natural rhythm. There is no chance for the interviewer to notice that the candidate misunderstood the question and needs a better prompt.

Some candidates are strong in conversation but uncomfortable in one-way recording.

Some candidates do not express themselves well on camera, but they perform very well in real work.

Some candidates come from cultures, homes, or family environments where recording personal video is not a simple matter.

If we ignore all of this, we risk confusing presentation with ability.

What AI Video Interviews May Actually Measure

AI video interviews can look objective because every candidate receives the same link and answers the same questions.

But sameness is not the same as fairness.

The format can introduce signals that are not directly related to the job:

• Is the candidate comfortable talking to a camera?
• Does the candidate have a private and quiet place to record?
• Is the internet connection stable?
• Is the lighting good?
• Is the background professional enough?
• Is the candidate fluent under recording pressure?
• Does the candidate's accent, expression, or nervousness affect the impression?

These signals may influence a human reviewer. They may also influence automated scoring if the system uses voice, expression, language patterns, or other behavioral indicators.

That is where the danger begins.

The employer may believe it is measuring potential.

In practice, it may be measuring how well the person performs inside a narrow digital format.

Privacy Is Not A Small Detail In Our Region

In Jordan and across the Middle East, privacy and culture matter.

For many women candidates, recording a personal interview video and storing it in a database is not a small administrative request. It can be a serious barrier.

Some families will not be comfortable with it. Some candidates will not be comfortable with it. Some will simply not apply, even if they are qualified.

If that happens, the company did not only create a digital process.

It created a filter that quietly pushes qualified people out.

That is why a video assessment should never be the only route to qualification.

If an employer wants to use video, there should be an equivalent alternative: a live interview, an audio-only option, a written assessment, a structured phone screen, an in-person route, or a human-reviewed work sample.

The alternative must be real, not a punishment.

Candidates Can Learn To Play The System

There is another problem: the more candidates understand how AI assessment works, the easier it becomes to prepare for the machine.

They can learn the expected keywords.

They can rehearse the structure.

They can adjust their tone, pacing, and examples.

They can answer in the way the system seems to reward, even if that answer does not reflect their real working ability.

This is not a moral judgment against candidates. Candidates adapt to the process they are given.

The problem is on the employer side: if the process rewards machine-facing performance, then the employer should not confuse that with deeper talent.

Recruitment AI Needs Human Ownership

AI in hiring should help humans make better decisions.

It can help prepare questions. It can summarize evidence. It can compare candidates against job-related criteria. It can flag missing information. It can help document why a candidate moved forward or did not.

But the employer must still own the decision.

The hiring team should know what the tool measures, what data it collects, how long that data is stored, who can access it, whether the candidate has an alternative, and how a human can review the result.

If those questions are not clear, the process is not ready.

Use AI to support better hiring decisions, not to hide weak process design behind a polished automated interview.

A Better Line

The better line is simple:

Use AI to support recruitment.

Do not force every candidate to perform for AI before a human sees them.

Do not make video the only path.

Do not treat camera comfort as proof of competence.

Do not treat privacy concerns as resistance to technology.

Do not allow an automated score to carry a decision that affects a person's career.

AI can help us build better hiring systems.

But a candidate is more than a recording.

And an AI video interview may not show who that candidate really is, or what they can actually do.

Sources

• U.S. EEOC and DOJ warning on AI/software in employment and disability discrimination risks
• EU AI Act Service Desk, Annex III: high-risk AI systems in employment, worker management, and access to self-employment
• EU AI Act Service Desk FAQ on prohibited practices including workplace emotion inference
• Jordan Personal Data Protection Law PDF from MODEE
• Basch & Melchers, 2019, applicant reactions to asynchronous video interviews
• NIST AI Risk Management Framework