Async Video Interviews Were Supposed to Scale Screening. AI Ghostwriters Just Broke Them.

AI ghostwriting for job interviews has gone from edge case to mainstream — and it's quietly dismantling one of the most widely adopted screening tools in the recruiter's stack. Async video interview platforms sold recruiters on a simple efficiency trade: replace scheduling overhead with standardized prompts, batch-evaluate recorded answers, and move only the best candidates forward. That trade made sense in 2022. It's stopped making sense now.

What Async Video Was Actually Supposed to Solve

The original value proposition was real. One-way video screening let a recruiter review 50 candidates in the time a phone-based process reached 10. Candidates recorded answers on their own schedule, recruiters evaluated asynchronously, and the bottleneck of mutual availability disappeared. On paper: same signal quality as a live phone screen, dramatically better throughput.

The assumption baked into that model was that the answers being recorded came from the candidates doing the recording. That assumption no longer holds.

What's Happening While the Camera Is On

Tools that generate real-time interview answers have been commercially available since late 2024. Some work as browser extensions overlaying suggested responses as the question plays. Others listen via microphone and generate text in a sidebar the camera doesn't capture. A third category refines whatever draft the candidate produces before the recording uploads.

The outcome is that a meaningful and growing share of async video answers don't represent how a candidate thinks or communicates under pressure. They represent how the candidate's AI tool handles interview prompts. Worse, because most of these tools draw on the same underlying models, the answers are converging. Recruiters are batch-evaluating polished, coherent responses that have become largely indistinguishable from one another — because they largely are.

This isn't speculation. It's the same dynamic that already played out with text resumes — the pattern is covered in detail in the post on AI-generated applications flooding your ATS. The resume screen collapsed first. The video screen is following the same path.

The Structural Problem Is Asynchrony Itself

Async video has the same vulnerability as written screening: it's not adaptive. The prompt is known in advance or inferable within seconds of the question playing. Most platforms allow retakes. There's no follow-up based on what the candidate actually says, no probe for a concrete example when an abstract one is given, no moment where an unexpected question disrupts a prepared script.

That makes it a content production environment, not a screening environment. The candidate — or their AI tool — has as much time as the platform allows, with full knowledge of the question. In those conditions, what gets assessed is not how someone thinks under spontaneous pressure. It's how well they optimize for the format.

Live conversation is structurally different. The latency of AI assistance in a real-time adaptive exchange is detectable — a pause to paste a question into a tool and retrieve an answer doesn't survive a conversation where someone is waiting for the next sentence. The cheating surface narrows sharply when the screening is live and the follow-ups are unpredictable.

How Asendia AI Approaches This

Asendia AI is a voice-first AI recruiter that conducts live screening conversations 24 hours a day, 7 days a week. When a candidate applies, Asendia calls them — that evening, that night, over the weekend — and runs a structured qualification conversation calibrated to the specific role. Not a prompt the candidate can pre-game with a browser extension. A real-time spoken exchange that adapts to what the candidate actually says.

If a candidate gives a vague answer, Asendia follows up for specifics. If they claim a particular skill, it probes for evidence in context rather than accepting the assertion and moving on. What comes back into your ATS is a qualification summary and verbatim conversation excerpts — signal from a live adaptive exchange, not a polished recording produced under unlimited revision conditions.

For agencies running campaigns where 400 applications arrive before the recruiter opens their Monday queue, Asendia absorbs that volume without adding headcount. Every candidate gets a real conversation within hours of applying. The shortlist the recruiter inherits is built from a live, adaptive screening format — one that's harder to systematically pre-game than anything async video offers, because the follow-up questions can't be anticipated before the call.

The platform connects directly to existing ATS workflows. No parallel system to manage. Same pipeline, now populated by candidates who've been spoken to in real time rather than by candidates who submitted optimized recordings under conditions that systematically favor AI assistance over authentic response.

Final Word

Candidates using AI tools to prep async video answers aren't doing something unusual. They're responding to a hiring environment that already uses AI to screen, rank, and filter them — the optimization runs in both directions now. The problem isn't the behavior. It's that async video was never designed to hold up against it. The screen doesn't need to be human to be reliable. It needs to be live, adaptive, and hard to pre-game. Async video, in its current form, is none of those things — and teams that haven't revisited their screening stack since 2022 are making decisions on a foundation that's significantly less solid than the platform's interface suggests.

Ready to transform your hiring strategy? Schedule a Demo with our founders today!