AI Interviews in Recruitment: Are They Fair, and How Does HR Stay in Charge?

First-round interviews are one of the most time-consuming parts of HR work. When hiring for many roles or branches, dozens or even hundreds of applicants need to be contacted, asked nearly the same questions, and written up. That is why more companies are looking at AI interviews: candidates answer structured questions, and the system prepares a transcript and an assessment.

A fair question follows: is this fair to candidates? And if AI is involved in scoring, who is actually making the decision? This article looks at both in a balanced way: where AI interviews can support fairness, where they can cause harm, and which controls need to be in place.

In short:

• Can be more consistent: every candidate gets the same core questions and the same scoring standard.
• Not automatically fair: a poor rubric, audio quality, accents, or disability can affect results.
• Candidates have a right to know: explain that AI is used, for what purpose, and how their data is handled.
• HR still decides: the AI score is input, not the final decision.

What Is an AI Interview in Recruitment?

An AI interview is an early-stage interview run by an AI system rather than directly by a recruiter. It can be text, video, or voice. Candidates answer a set of structured questions, and the system can ask follow-up questions based on their answers.

The system then produces a transcript, a summary, and an assessment against defined criteria, which the recruiter uses to decide who moves on to the hiring manager.

One point matters from the start: an AI interview changes how the first round is run, not who is responsible for choosing candidates. That distinction determines whether the system is used fairly.

Where AI Interviews Can Be Fairer

Manual interviews are not fully neutral either. A recruiter who has called dozens of candidates in one day will understandably get tired. The order of questions shifts, some candidates get more time than others, and first impressions often shape the notes.

A well-designed AI interview can reduce some of these problems:

• The same questions for everyone. Every candidate for the same role gets the same core questions.
• The same scoring standard. Answers are assessed against the same rubric, not the assessor's mood.
• Flexible timing. Candidates who are currently employed can answer outside office hours without matching a recruiter's schedule.
• Everything is recorded. A transcript lets others review the interview instead of relying on one person's brief notes.

This matches structured interviewing, a long-recommended hiring practice. But consistency does not guarantee fairness. A standard that is the same for everyone but wrong will disadvantage every candidate in the same way.

Where the Risks Are

AI interviews also bring new risks, and some are not obvious at first. The table below summarizes the ones that deserve the most attention.

Adaptive follow-up questions also mean two candidates do not always answer exactly the same questions. Scoring should rest on the core questions and the rubric, not on the length or style of an answer.

Candidate Transparency and Consent

Candidates have a right to know they are being interviewed by AI. Hiding it is an ethical problem, and candidates who feel misled lose trust in the company as an employer.

Before the interview starts, explain briefly and clearly:

• That this stage is run by AI and is recorded.
• What data is collected: voice recordings, transcripts, and assessments.
• What the data is used for, and who sees the results.
• That the decision is made by the HR team, not the AI.
• How long the data is kept, and how to contact the company with questions.
• What options exist if the candidate needs an adjustment or a different format.

In Indonesia, processing candidate data falls under Law No. 27 of 2022 on Personal Data Protection (the PDP Law). The law requires a lawful basis for processing personal data, such as explicit consent, and gives data subjects the right to information about the purpose of processing. It also gives individuals the right to object to decisions based solely on automated processing that significantly affect them. Whether someone passes a hiring stage can fall into that category, so human involvement in the decision is more than good practice. Confirm the details of how this applies with your company's legal team.

The Role of HR: AI Assesses, People Decide

A healthy division of roles is simple: AI handles repetitive work, HR owns the decision. In practice, a good review flow looks like this:

1. Read the summary and the score per criterion, not just the total score.
2. Open the transcript for candidates near the threshold, or where the result looks odd.
3. Check the context, such as notes on audio quality or answers that were cut off.
4. Decide and record the reason, especially when the decision differs from the score.
5. Move to the next stage, the interview with the hiring manager, which is still run by people.

Recruiters also need to feel free to disagree with the score. If HR always follows the number without reading the transcript, the decision has in practice been handed to the machine, even if a person signs off on paper.

Documentation and Data Protection

One advantage of AI interviews is the documentation trail. Transcripts, scorecards, and decision notes make the process reviewable, both for internal evaluation and when a candidate questions the outcome.

But complete documentation also means more personal data. A few principles to hold on to:

• Collect only what you need. Ask about things relevant to the role. Avoid questions about religion, marital status, or other personal matters unrelated to the job.
• Limit access. Only the team handling that opening should be able to open recordings and transcripts.
• Set a retention period. Delete data for candidates who do not move forward after a period that has been set and communicated.
• Provide a request channel. Candidates need to know how to request access to, correction of, or deletion of their data.
• Assess the impact. The PDP Law requires a data protection impact assessment for high-risk processing, including automated decision-making with significant effects. Involve your legal team or data protection officer from the start.

How to Introduce AI Interviews Fairly

If your company decides to try AI interviews, start small:

1. Pick one high-volume role with clear criteria, such as operational or customer service staff.
2. Build the rubric with the hiring manager. Every criterion should have a clear link to the job.
3. Test the questions internally before using them with real candidates.
4. Prepare the notice and an alternative route for candidates who need one.
5. Start with full manual review. A recruiter reviews every result before an automatic score threshold is considered.
6. Compare with the old process. Are the candidates who pass of comparable quality? Is any group being filtered out without a clear reason?
7. Expand gradually once the rubric, the notice, and the review flow have proven they work.

How Humax Handles This

Humax, the recruitment platform from IOI, offers AI voice interviews for the first round. Every candidate gets the same questions and the same standard, and every interview produces a transcript and a scorecard. Answers are scored per criterion against the rubric the team sets for the role, so the company decides what matters.

The Humax candidate pipeline uses configurable gates: all candidates advance, candidates advance above a score threshold, or a recruiter reviews them manually. This lets teams start with full manual review, as suggested above. Humax also offers CV screening and smart search, and connects to Google Calendar, Microsoft Teams, and WhatsApp for interview scheduling and candidate updates.

The principle stays the same: the AI in Humax does not make hiring decisions. The HR team reviews the results and decides who moves forward. Candidate notices, retention policies, and impact assessments remain the responsibility of the company as the data controller.

Frequently Asked Questions

Do AI interviews replace recruiters?

No. AI interviews take over the repetitive first round. Recruiters still review the results, decide who moves forward, and run the later interview stages.

Do candidates need to be told their interview is run by AI?

Yes. Beyond trust, the PDP Law requires that data subjects be informed about the purpose of processing and how their data is handled.

What if a candidate cannot take a voice interview?

Offer an alternative, such as a live interview with a recruiter, without reducing that candidate's chances.

Can AI scores be used to reject candidates automatically?

Technically yes, but they should not be the only basis. Review candidates close to the threshold, and make sure candidates have a way to raise an objection.

What data should be kept?

Transcripts, scorecards, and decision notes, for as long as needed for recruitment and evaluation, in line with the retention period communicated to candidates.

Conclusion

AI interviews can make early-stage hiring more consistent and better documented than rushed phone screens. But fairness does not come from the technology itself. It comes from a relevant rubric, informed candidates, an available alternative, and an HR team that actually reads the results before deciding.

Use AI to give every candidate the same chance to be heard, and keep the decision in human hands.

Want to see how it works? Read about Humax on the IOI site, or try it at humax.id.