How common are AI-conducted interviews in 2026?

According to Greenhouse's 2026 Candidate AI Interview Report, surveying 2,950 job seekers, 63% have now experienced an AI interview, up 13 percentage points from just six months earlier.

Will I even be told an AI is interviewing me?

Often not upfront. Greenhouse's survey found 70% of candidates weren't clearly told AI would be involved before the interview started, and one in five only found out after it had already begun.

What actually helps in an AI-conducted interview?

Speak in complete, specific sentences the way you would to a person, since these systems are typically scoring transcribed speech for keyword and structure signals, not tone. Name concrete tools, numbers, and outcomes explicitly rather than assuming context carries over, and treat the recorded format as literally as a written application, since a large share of these systems allow no human review before a rejection goes out.

Blog

AI Is Now Interviewing You First

Q: Is there real evidence AI interviews are biased?

Yes. A Stanford HAI study analyzing 3.4 million job applications found AI hiring systems discriminated against 26% of Black applicants and 15% of Asian applicants, and that fixing the disparity would have advanced roughly 40,000 more applications to the next round.

December 8, 2026

If you applied to a large bank, a consulting firm, or a big tech company in the last year, there is a real chance your first interview was not with a person. It was a recorded video or chat session, scored by a model, before anyone at the company saw it. That is no longer an edge case. It is close to the default.

How common this actually is now

Greenhouse's 2026 Candidate AI Interview Report, based on a survey of 2,950 active job seekers (1,200 of them in the US), found that 63% have now been interviewed by AI, a 13 point jump from just six months before the survey. This is not a slow trend, it is accelerating inside a single year.

The candidate experience of that shift is rough. 38% of respondents said they abandoned a hiring process specifically because it included an AI interview, and another 12% said they would if it came up again. Roughly half of everyone who completed an AI interview reported hearing nothing back afterward at all, not even a rejection.

You often will not be told upfront

The same Greenhouse survey found that 70% of candidates were never clearly told AI would be involved before their interview started. One in five said they only realized it once the interview had already begun. If you have had an oddly stilted, one way video interview recently and could not tell if a person was on the other end, there is a real chance one was not.

The bias research is real, not speculative

This is not just an inconvenience. A Stanford HAI study led by Rishi Bommasani, Sarah Bana, Kathleen Creel, Dan Jurafsky, and Percy Liang analyzed 3.4 million people submitting 4 million applications to 1,700 job postings across 150 employers, all scored by a single third party hiring algorithm. Using the EEOC's own four fifths rule for adverse impact, the researchers found the system discriminated against 26% of Black applicants and 15% of Asian applicants relative to the most favored group. Had the algorithm treated candidates evenly, the researchers estimate roughly 40,000 more applications from those groups would have advanced to the next stage.

The study also documented a compounding effect the researchers called systemic rejection: among candidates who submitted four applications through this same system, 10% were rejected everywhere they applied. When many employers run the same algorithmic screen, one system's blind spot becomes every one of those employers' blind spot at once, for the same candidate.

What actually helps, given all of this

None of this means an AI interview is unbeatable, it means it needs to be treated differently than a conversation with a person. A few things that follow directly from how these systems are built:

Answer in complete, specific sentences. Most of these systems score a transcription of what you said, not your tone or presence. A vague or fragmented answer that a human interviewer might mentally fill in the gaps for reads as genuinely thin to a system parsing text.
Name the concrete thing, do not assume context carries over. The tool, the number, the outcome, stated explicitly, the same discipline that matters for a resume passing an ATS applies here for the same underlying reason: automated systems reward what is stated plainly, not what is implied.
Treat it as literally as a written application. With around half of respondents in the Greenhouse survey reporting no follow up communication at all after an AI interview, and many systems allowing rejection with no human review, there is often no second chance to clarify an answer the way there might be with a person reading your tone and intent in real time.

The honest caveat

None of this is a guarantee, and none of the research above claims these systems can be reliably outsmarted. What it does establish is that this is now common enough, and documented enough, that dismissing it as a fringe experience is no longer accurate. If you want to practice the format itself, ResuMakeAi's AI interview prep generates likely questions with model answers from your actual resume and a specific job posting, the same resume and posting language worth having ready before any interview, automated or not. See common interview questions and how to answer them for the content side of preparing, independent of who or what is asking.