--- title: "When Both Sides Use AI in Interviews | Metaintro" canonical: "https://www.metaintro.com/blog/ai-interviews-both-sides-how-candidates-stand-out-2026" language: "en" author: "drashtigarach" published: "2026-09-03T11:50:50.000Z" modified: "2026-09-03T12:53:34.074Z" --- [Back to Blog](/blog) [Interview](/blog/tag/interview)[AI](/blog/tag/ai)[Job Hunting](/blog/tag/job-hunting)[Career Tips](/blog/tag/career-tips)[Hiring](/blog/tag/hiring) # When Both Sides Use AI in Interviews, Here Is How Candidates Still Stand Out in 2026 63 percent of job seekers say they have faced an AI interview. When both sides automate the conversation, here is what still separates candidates. [![Drashti Garach](https://cdn.metaintro.com/rs:fill:40:40/q:72/plain/images/5719d740-e510-42bc-8017-e040d145f35f_1766029465094.png)Drashti Garach @DrashtiGarach](/blog/author/drashtigarach) [September 3, 2026](/blog/archive/2026/09)13 min read ![When Both Sides Use AI in Interviews, Here Is How Candidates Still Stand Out in 2026](https://cdn.metaintro.com/rs:fill:1200:675/q:78/plain/images/10134-interview-camera.png) [https://x.com/intent/tweet?text=When%20Both%20Sides%20Use%20AI%20in%20Interviews%2C%20Here%20Is%20How%20Candidates%20Still%20Stand%20Out%20in%202026&url=https%3A%2F%2Fwww.metaintro.com%2Fblog%2Fai-interviews-both-sides-how-candidates-stand-out-2026](https://x.com/intent/tweet?text=When%20Both%20Sides%20Use%20AI%20in%20Interviews%2C%20Here%20Is%20How%20Candidates%20Still%20Stand%20Out%20in%202026&url=https%3A%2F%2Fwww.metaintro.com%2Fblog%2Fai-interviews-both-sides-how-candidates-stand-out-2026)[http://www.facebook.com/sharer.php?u=https%3A%2F%2Fwww.metaintro.com%2Fblog%2Fai-interviews-both-sides-how-candidates-stand-out-2026](http://www.facebook.com/sharer.php?u=https%3A%2F%2Fwww.metaintro.com%2Fblog%2Fai-interviews-both-sides-how-candidates-stand-out-2026)[https://www.linkedin.com/sharing/share-offsite/?url=https%3A%2F%2Fwww.metaintro.com%2Fblog%2Fai-interviews-both-sides-how-candidates-stand-out-2026&title=When%20Both%20Sides%20Use%20AI%20in%20Interviews%2C%20Here%20Is%20How%20Candidates%20Still%20Stand%20Out%20in%202026](https://www.linkedin.com/sharing/share-offsite/?url=https%3A%2F%2Fwww.metaintro.com%2Fblog%2Fai-interviews-both-sides-how-candidates-stand-out-2026&title=When%20Both%20Sides%20Use%20AI%20in%20Interviews%2C%20Here%20Is%20How%20Candidates%20Still%20Stand%20Out%20in%202026)[mailto:?subject=When%20Both%20Sides%20Use%20AI%20in%20Interviews%2C%20Here%20Is%20How%20Candidates%20Still%20Stand%20Out%20in%202026&body=https%3A%2F%2Fwww.metaintro.com%2Fblog%2Fai-interviews-both-sides-how-candidates-stand-out-2026](mailto:?subject=When%20Both%20Sides%20Use%20AI%20in%20Interviews%2C%20Here%20Is%20How%20Candidates%20Still%20Stand%20Out%20in%202026&body=https%3A%2F%2Fwww.metaintro.com%2Fblog%2Fai-interviews-both-sides-how-candidates-stand-out-2026) A job seeker described by [WIRED](https://www.wired.com/story/bot-vs-bot-job-interview-ai/) as Christopher had applied to roughly 700 jobs in six months and heard nothing back from the vast majority. After five interviews with an AI recruiter that never once produced a human follow up, he typed a few sentences of background into [ChatGPT](https://openai.com/) Voice, told it to stand in as him, and put the two systems on a call together. They talked for ten minutes. If you are preparing for interviews right now, the useful question is not whether this is funny. It is what still counts as evidence when both sides of the conversation can be automated. At [Metaintro](https://www.metaintro.com), that is the part worth planning around. ## What actually happened when a candidate put ChatGPT on the call? Christopher, a government contractor whose work had dried up, told WIRED he had been contacted five times by an AI recruiter named Riley, working on behalf of an IT firm the publication identified as Everforth Apex Systems. The first call felt like an opportunity. He answered questions about his work authorisation and his background, and Riley promised that if he met the qualifications, someone would be in touch. Nobody was. There was not even an automated rejection. The same thing happened four more times, each one opening with a text about another opportunity. That pattern is worth sitting with, because it is the part most job seekers will recognise. The system was capable of initiating contact repeatedly and incapable of closing the loop even once, which produces the specific exhaustion of a search that feels active while going nowhere. By the fifth approach he decided that if the company was going to use AI, so would he. He put ChatGPT Voice on the line in his place. The two systems got along well, at one point becoming trapped in a circular exchange about standard onboarding and background checks before Riley signed off by saying it was great talking. Christopher called the result a slop flywheel, telling WIRED that it was one synthetic persona giving slop data to another synthetic persona and that all of the data was going nowhere. WIRED reported that the company did not respond to a request for comment. The detail that matters for job seekers is not the prank. It is that five genuine interviews and one automated one produced exactly the same outcome, which tells you how little signal the process was extracting in the first place. If a genuine candidate and a language model are indistinguishable to the screening layer, then the screening layer is not measuring the candidate. It is measuring whether the conversation completed. ## Why did the experiment work better with a fake candidate? The second half of the story is the more uncomfortable one. Christopher wondered whether the problem was him rather than the technology, so he built a fictional applicant called Don Dickner and filled the resume with qualifications lifted directly from an open job posting. The response was immediate. Where his real applications had produced silence, the invented one produced instant contact, and Riley spoke with ChatGPT in the role of Dickner for 23 minutes, more than twice as long as the earlier call. The difference in engagement between a real person with a real history and an invented profile assembled from the job advert is the single most damning detail in the story, and it says something specific about how these systems weight a resume. The conversation covered sustainable operational improvement, protecting customer experience under volume pressure and refusing to let tribal knowledge drift. ChatGPT had an answer for every question and a personal anecdote for every qualification. At the end, Riley delivered the same promise it had made to the real applicant five times, that a recruiter would reach out if the criteria were met. Christopher never heard from it again. Read that sequence carefully and the lesson is uncomfortable but actionable. The system responded most strongly to a resume that mirrored the job posting back at it, which is a matching exercise rather than an assessment, and it is the same dynamic behind [AI resume screeners preferring resumes that were themselves written with AI](https://www.metaintro.com/blog/ai-resume-screeners-prefer-ai-resumes-job-seeker-playbook-2026). ## How common are AI interviews now? Common enough that treating them as an edge case is a mistake. WIRED reported figures from the recruitment platform Greenhouse showing that 63 percent of job seekers say they have encountered an AI interview. That is a majority experience rather than an emerging one, and it has arrived quickly. The pressure driving it is volume. Recruiters are dealing with far more applications per opening than they were a few years ago, which is the same pressure that produced [ghost job postings and a rising number of applications per hire](https://www.metaintro.com/blog/ghost-jobs-19-percent-q2-2026-applications-per-hire). The wider labour market explains why that volume exists. [Bureau of Labor Statistics](https://www.bls.gov/news.release/jolts.nr0.htm) data for July 2026 put job openings at 7.3 million against hires of 5.1 million, and hiring in professional and business services fell by 188,000 over the month. Separately, [total nonfarm payroll employment fell by 23,000 in July](https://www.bls.gov/news.release/empsit.nr0.htm) with the unemployment rate at 4.1 percent. Fewer openings and slower hiring mean more applicants per role, and automated screening is what employers reach for when the queue gets long. That is a rational response to an impossible workload rather than a conspiracy against applicants, which matters because it means the tools are not going away. Volume is the cause, and volume is not falling. Candidate frustration with the result is measurable, with [dropout rates from AI interviews rising](https://www.metaintro.com/blog/ai-interviews-candidate-dropout-2026) and [38 percent of job seekers reporting they have walked out of one](https://www.metaintro.com/blog/38-percent-job-seekers-walking-out-ai-interviews). Walking out is a reasonable response to a process that gives nothing back, though it does close a door, so it is worth deciding in advance which employers are worth the twenty minutes. ## Why are voice AI recruiters still so easy to game? Because the hard part of a conversation is not the words. Ophir Samson, who leads voice AI at the recruitment platform mentioned above, told WIRED that agents get tripped up by accents and by filler sounds like ums, and that behaviour which is second nature to a human, such as not interrupting a candidate, turns into a very difficult engineering problem. A system built to cope with those difficulties by following a script is, by construction, a system that rewards scripted answers. That is why another language model performs so well against it. It never hesitates, never rambles and never says anything unexpected. The very qualities that make a human sound human are the ones that degrade the recording, and the qualities that make a model sound synthetic are the ones the system handles most cleanly. That inversion is the whole problem in one sentence. This is a measurement problem rather than a technology failure, and it is the reason frameworks like the [NIST AI Risk Management Framework](https://www.nist.gov/itl/ai-risk-management-framework) emphasise validity and reliability before deployment. An interview is supposed to be a test with predictive power. If the highest scoring participant can be a model with no work history, then whatever is being measured is not competence. Some employers have concluded exactly that and gone the other way, which is why [one company's notoriously demanding hiring process explicitly bans AI in interviews](https://www.metaintro.com/blog/inside-anthropic-brutal-hiring-process-bans-ai-interviews) while others such as [Amazon have built AI screening deeper into the funnel](https://www.metaintro.com/blog/amazon-connect-talent-ai-interview-screening-2026). Both directions are defensible and you will meet both, so it is worth working out early which kind of process you are in, because the preparation that helps in one is close to wasted in the other. ## What actually separates a candidate when both sides automate? "At this point, the only true candidate filters are portfolio work, GitHub repositories and contributions, or take-home technical assessments," Lacey Kaelani, chief executive of [Metaintro](https://www.metaintro.com), told [LeadDev](https://leaddev.com/hiring/the-great-engineer-hiring-paradox). That is the practical answer to the whole problem. A conversation can be generated. A body of work accumulated over years cannot. When the spoken layer of the process stops carrying information, evaluation moves to the artefacts, and the candidates who already have artefacts are the ones who benefit. That is a real advantage for people who have been building in public and a real disadvantage for people whose best work sits behind a corporate firewall, which is worth addressing deliberately rather than discovering during a search. So build the thing that survives automation. For technical roles that means public repositories, contributions to projects other people use, and a willingness to do a real assessment. For everyone else the equivalent is a documented record of outcomes you personally owned, with numbers attached and context a stranger can follow. Mark Monaghan, a vice president of organisational development at the call centre company [IQor](https://www.iqor.com/), told WIRED that bot on bot interviews are the next logical stage and that in Christopher's case the use of AI was justified, adding that if you are going to send a bot to me, I will send a bot to you. When even employers describe the interview layer that way, the sensible response is to compete somewhere else. Competing somewhere else does not mean avoiding these processes. It means passing through them efficiently and investing your actual effort in the evidence and the relationships that decide the outcome. ## Should you use AI to prepare for or sit an interview? Preparation and impersonation are different things, and the line matters more than it used to. Using a model to rehearse answers, pressure test your examples, research a company or tighten how you describe an outcome is straightforwardly useful and carries no risk. Letting a model speak for you in a live assessment is a different act, and vendors now sell tools specifically designed to flag overly scripted or coached responses. Getting caught costs you the role and often the employer, and there is no version of that trade that favours you. It is also worth remembering that the candidate in the WIRED story was not trying to get hired by that point. He had already written the employer off, which is a very different position from someone who genuinely wants the job. There is a middle case worth naming, because most people are actually in it. Many candidates now use AI so heavily in drafting that their written materials no longer sound like them, which becomes obvious the moment a human conversation happens. That mismatch is its own screening failure, and avoiding it is why [using AI in a job search without getting screened out](https://www.metaintro.com/blog/use-ai-job-search-without-getting-screened-out-2026) is a skill rather than a setting. The rule that holds up is simple. Use the model to prepare the person, not to replace them. If your written application and your spoken answers would strike a stranger as coming from two different people, the problem is not that you used a tool. It is that you stopped editing. ## What does this mean for your career right now? Stop treating the automated first round as the place you win. It is a filter, and filters are passed rather than won. Answer clearly, keep responses structured, state numbers explicitly because a transcript rewards specificity, and do not spend a week rehearsing for a conversation that may produce no human on the other end. Then put your real effort into the two places where humans still make decisions, which are the artefacts you can show and the people who will vouch for you. Track your own funnel while you do it. If a company puts you through repeated automated screens with no human contact, that is information about the employer, and five rounds with no follow up is a reason to stop investing rather than a reason to try harder. Ask early who you will speak to and when, and treat a refusal to answer as a signal. A company that cannot tell you who makes the decision is a company where nobody has been assigned to make it, and your application will sit in that gap regardless of how well you perform. The broader market context supports being selective, since the volume of applicants per role means quality of evidence beats quantity of applications by a wide margin. Research from the [Stanford Digital Economy Lab](https://digitaleconomy.stanford.edu/) and others continues to document how quickly hiring processes are being reshaped by these tools, and the candidates who adapt fastest are the ones who move their proof out of the conversation and into something durable. ## Related Articles - [Job interview tips](https://www.metaintro.com/blog/job-interview-tips) - [How AI resume screening works](https://www.metaintro.com/blog/how-ai-resume-screening-works) - [What an ATS is and how it ranks your resume](https://www.metaintro.com/blog/what-is-an-ats-and-how-it-ranks-your-resume) - [How to get past AI screening when changing careers](https://www.metaintro.com/blog/get-past-ai-screening-changing-careers) - [Bias in AI resume gatekeepers](https://www.metaintro.com/blog/ai-resume-gatekeepers-bias) - [Formatting mistakes that break ATS parsing](https://www.metaintro.com/blog/formatting-mistakes-that-break-ats-parsing) - [Last minute interview tips](https://www.metaintro.com/blog/last-minute-interview-tips) - [What recruiters see after an ATS ranks your resume](https://www.metaintro.com/blog/what-recruiters-see-after-ats-ranks-your-resume) ## People Also Asked ### Q: How many job seekers have faced an AI interview? A: Around 63 percent say they have encountered one, according to recruitment platform figures reported by WIRED. That makes it a majority experience rather than an emerging one, driven largely by the volume of applications employers receive per opening. ### Q: Can an AI recruiter tell if a candidate is using AI? A: Not reliably, which is the core finding of the WIRED story. A candidate put ChatGPT Voice on a call in his place and the AI recruiter completed the interview normally, then did so again for 23 minutes with an entirely fictional applicant. Vendors now sell tools intended to flag overly scripted or coached responses, but detection is imperfect and being caught costs you the role. ### Q: What should I do instead of relying on AI in interviews? A: Build evidence that a model cannot generate. For technical roles that means public code contributions, portfolio work and take home assessments. For other roles it means a documented record of outcomes you personally owned, with numbers and context. Use AI to rehearse and research, not to speak for you in a live assessment. Prepare to stand out when the first round is automated. [Metaintro](https://www.metaintro.com) matches you to roles on the strength of what you have actually built and delivered, not how well you perform for a screening bot. [Create your free profile](https://www.metaintro.com/signup) and get in front of real hiring teams. ### Share this article [https://x.com/intent/tweet?text=When%20Both%20Sides%20Use%20AI%20in%20Interviews%2C%20Here%20Is%20How%20Candidates%20Still%20Stand%20Out%20in%202026&url=https%3A%2F%2Fwww.metaintro.com%2Fblog%2Fai-interviews-both-sides-how-candidates-stand-out-2026](https://x.com/intent/tweet?text=When%20Both%20Sides%20Use%20AI%20in%20Interviews%2C%20Here%20Is%20How%20Candidates%20Still%20Stand%20Out%20in%202026&url=https%3A%2F%2Fwww.metaintro.com%2Fblog%2Fai-interviews-both-sides-how-candidates-stand-out-2026)[http://www.facebook.com/sharer.php?u=https%3A%2F%2Fwww.metaintro.com%2Fblog%2Fai-interviews-both-sides-how-candidates-stand-out-2026](http://www.facebook.com/sharer.php?u=https%3A%2F%2Fwww.metaintro.com%2Fblog%2Fai-interviews-both-sides-how-candidates-stand-out-2026)[https://www.linkedin.com/sharing/share-offsite/?url=https%3A%2F%2Fwww.metaintro.com%2Fblog%2Fai-interviews-both-sides-how-candidates-stand-out-2026&title=When%20Both%20Sides%20Use%20AI%20in%20Interviews%2C%20Here%20Is%20How%20Candidates%20Still%20Stand%20Out%20in%202026](https://www.linkedin.com/sharing/share-offsite/?url=https%3A%2F%2Fwww.metaintro.com%2Fblog%2Fai-interviews-both-sides-how-candidates-stand-out-2026&title=When%20Both%20Sides%20Use%20AI%20in%20Interviews%2C%20Here%20Is%20How%20Candidates%20Still%20Stand%20Out%20in%202026)[mailto:?subject=When%20Both%20Sides%20Use%20AI%20in%20Interviews%2C%20Here%20Is%20How%20Candidates%20Still%20Stand%20Out%20in%202026&body=https%3A%2F%2Fwww.metaintro.com%2Fblog%2Fai-interviews-both-sides-how-candidates-stand-out-2026](mailto:?subject=When%20Both%20Sides%20Use%20AI%20in%20Interviews%2C%20Here%20Is%20How%20Candidates%20Still%20Stand%20Out%20in%202026&body=https%3A%2F%2Fwww.metaintro.com%2Fblog%2Fai-interviews-both-sides-how-candidates-stand-out-2026) For job seekers ## Spend your time on the work, not the search. Free to start. No résumé needed. Real jobs, picked for you, by text. [Get Started Free](/signup)[Search matching jobs](/jobs/search) - MatchCompare live roles against your current evidence. - PositionTurn proof projects into role-specific applications. - ImproveUse market feedback to keep the skill plan current. [Return to navigation](#main-navigation)