The job interview has always been a performance staged on someone else’s terms: the employer’s office, the employer’s schedule, the employer’s pace. AI is changing that arrangement in ways that are genuinely new, and genuinely uncomfortable. The late-night job interview is now a measurable phenomenon, and understanding it means looking past the novelty to see what it reveals about who the modern hiring process was never designed to serve.
The Candidate Who Can Only Interview at Midnight
Tim Millard’s experience is a useful entry point. Six months into a job search, the Atlanta-based communications professional received an invitation to interview for a media relations role. He expected a human on the other end. Instead, he found himself alone in his tidied living room at 9:30 pm, watching questions appear on a screen, waiting for a timer to count down before he could speak. Thirty seconds to think. Two minutes to answer. A “ding” to mark the transition. Months later, an automated rejection arrived.
What Millard experienced was an asynchronous AI video interview: a format that records candidates answering pre-set questions, then passes the footage and a rubric-based score to a recruiter. The format is spreading. And one of its most telling side effects is when people choose to complete it.
The Numbers Behind the Late-Night Shift
Arsham Ghahramani cofounded Ribbon, a company that builds voice-AI recruitment software. Drawing on data from more than 500 companies using the platform, Ribbon found that 24% of its AI interviews take place between 10 pm and 2 am local time. For manufacturing clients specifically, that figure rises to 35%.
Greenhouse, a separate hiring platform, reports that roughly 15 to 20% of candidates who interact with its voice agent choose to schedule their interview at night.
These numbers are not a sign of candidates burning the midnight oil out of ambition. They reflect a structural reality: many people who are actively job searching cannot step away from their current lives during business hours. Hourly workers cannot leave a shift. Parents cannot find 30 uninterrupted minutes until the household quiets down. Factory floor workers cannot conduct a voice interview over the noise of machinery. For these candidates, the flexibility of an AI interview is not a perk. It is the only viable window.
This is what most coverage of AI hiring tools misses. The conversation tends to focus on efficiency gains for employers, or on the strangeness of talking to a machine. The more significant shift is that asynchronous AI interviews are, for the first time, making the process accessible to people who were structurally excluded from it before.
Skepticism, Scoring, and the Black Hole Problem
Accessibility does not equal acceptance. A Greenhouse survey found that nearly two-thirds of job applicants have been interviewed by an AI agent, a figure that rose 13 percentage points over just six months. Yet 38% of American candidates told Greenhouse they had withdrawn from a hiring process rather than complete an AI interview. Another 12% said they would drop out if one were required.
The anxiety is not irrational. Candidates often do not know how their footage will be used, who will see it, or what criteria the AI is applying. The scoring rubrics vary by role: a welder might be evaluated on technical certifications, while a sales candidate might receive an “enthusiasm” score. AI analyzes the footage against the rubric and assigns a number. That number goes to a recruiter. The footage, in many cases, goes unwatched by any human.
Ophir Samson, who leads Greenhouse’s voice AI division, frames the AI interview not as a replacement for human judgment but as a layer added before it. His argument is that the alternative for most candidates is not a human interview. It is silence. Résumés disappear into applicant tracking systems. Qualified people are indistinguishable from the volume of AI-generated applications surrounding them. The AI interview, in this framing, is a mechanism for surfacing candidates who would otherwise never reach a human recruiter at all.
That framing does not resolve the transparency problem. But it reframes the question: the choice is less often “AI interview versus human interview” and more often “AI interview versus no response at all.”
What This Moment Actually Reveals
The late-night interview trend is a symptom of a hiring system under pressure from multiple directions simultaneously. Employers are processing more applications than ever, many of them AI-generated. Candidates are competing in a market where differentiation is increasingly difficult. And the tools being deployed to manage this volume are outpacing the norms and expectations that candidates bring to the process.
Millard’s story ends without resolution. A year after losing his job, he is still searching. The experience was disorienting enough that he turned it into a one-man show, which debuted at the Atlanta Fringe Festival. His description of the job search as “screaming into the void” captures something precise: the asymmetry between the effort candidates invest and the feedback they receive.
AI interviews do not fix that asymmetry. They redistribute it. The convenience is real for some candidates. The opacity is real for all of them.
In Short
AI-powered and asynchronous video interviews are enabling a measurable shift in when people job search, with a significant share of interviews now happening late at night. This flexibility genuinely expands access for candidates who cannot interview during business hours. At the same time, candidate skepticism is high: a substantial portion of applicants are withdrawing from processes that require AI interviews, largely because of uncertainty about how they are evaluated. The deeper issue is not the technology itself but the conditions that made it necessary: a hiring environment where volume has overwhelmed human capacity to respond, and where candidates often receive no feedback at all. AI interviews are one response to that breakdown. They are not a resolution of it.
Based on reporting from Wired.