“I’m Daniel Philip Johnson. I’ve been called the songbird of my generation.”
— Me, quoting Step Brothers while psyching myself up to talk to a webcam.
Okay — not really. But I have spent a shocking amount of time talking to myself in front of a camera, trying to impress an algorithm. That’s why I had such a great time using HireVue — the AI-powered interview platform that removes one unnecessary part of the hiring process: other people.
If you’re unfamiliar, HireVue is what happens when someone looks at a regular job interview and thinks, “What if we made this colder, lonelier, and judged entirely by a machine?” You get a link, some instructions, and then you’re off — recording answers to pre-set questions while staring into your webcam. No interviewer. No feedback. Just you and a timer.
Interviewing Into the Void
At first, I thought: how bad can it be? I mean, I’ve done video calls. I’m comfortable talking. I’m reasonably tech-savvy.
But there’s something deeply weird about smiling at your own reflection while trying to project confidence and warmth — to no one. You’re expected to maintain eye contact, speak clearly, smile naturally. And you do. Because that’s what we’ve trained ourselves to do in real conversations. You nod. You pause for reactions. You try to build rapport.
And then it hits you: you’re faking a connection with a void. Obviously platonic, but still a connection. It’s like flirting with your microwave. No matter how well you do, it won’t tell you if it enjoyed the chat.
No Feedback Loop, No Humanity
In a real interview, there’s a feedback loop. You say something, and maybe it doesn’t quite land — so the interviewer asks, “Can you clarify?” or “I didn’t quite follow.” That little nudge helps you reframe your point, recover, even shine.
HireVue doesn’t give you that. It just lets you bumble on. There’s no “wait, let me try that another way.” No chance to read the room — because there is no room. Just a cold, unblinking camera and a countdown timer.
It’s like doing a whiteboard session where the interviewer says nothing. You draw, you talk, you explain — and all the while, you're being silently observed. Judged.
“There’s nothing like trying to build rapport with a loading icon.”
Honestly? It felt like being in a game of Portal. You're the test subject. The algorithm is GLaDOS. You keep talking, hoping you're getting it right, with no idea what “right” even means.
At one point, I genuinely felt like I’d been taken captive and was recording a ransom plea.
“Please, let me work for your company. I have marketable skills. I can meet KPIs. Think of my family.”
Because who knows — maybe if I looked desperate enough, the algorithm would finally believe my response.
The Black Box with a Scorecard
It’s easy to laugh at the absurdity of AI interviews — until you realise what’s really happening.
Behind the scenes, algorithms are evaluating your performance. Supposedly looking for things like communication skills, enthusiasm, and problem-solving ability. But how? Based on what? Says who — the algorithm?
This is the heart of the “black box” problem in AI hiring. HireVue, like many AI interview platforms, does not provide a transparent breakdown of how specific non-verbal cues — facial expressions, vocal tone, body language — are weighted. You’re assessed by unseen metrics, against undefined standards, with no opportunity to ask for clarification or feedback.
“Getting feedback from an algorithm is like yelling into a canyon and hoping it explains your echo.”
It’s not just frustrating. It’s dangerous.
As a Forbes Tech Council article highlights, "In some cases, deeper AI learning models have become so advanced that even their creators don't fully comprehend how they work" [1]. That’s a terrifying sentence — especially when your ability to land a job might hinge on whether or not you blinked too much.
“If the machine doesn’t know why it said no — how are we supposed to learn from it?”
These systems are trained on historical data. And if that data reflects existing human bias — around gender, race, age, or accent — the system doesn’t remove bias. It scales it [2]. Amazon learned that the hard way with its now-retired AI hiring tool, which penalised CVs containing the word “women’s” [2]. HireVue, for its part, publicly stopped using facial analysis after backlash [3]. But without full transparency into how their Natural Language Processing (NLP), sentiment scoring, or behavioural analytics work — candidates are left guessing.
What should be a conversation becomes a monologue — with no red pen, no context, no second chance. It’s the digital version of “Computer says no” — except this time, it’s your future on the line.
Do They Know It’s Biased?
To be fair, HireVue and similar companies don’t pretend bias doesn’t exist. They claim to address it — and to their credit, they’ve moved beyond internal promises.
They’ve brought in independent firms like DCI Consulting Group [4], ORCAA [5], and Landers Workforce Science LLC [6] to audit their systems. These are not subsidiaries. They’re not internal rebrands wearing a new logo and a trench coat.
But let’s be honest: if I pay someone to evaluate me, I’m at least hoping for a good review — or at the very least, a polite one.
“You don’t hire the food critic to write the menu.”
Would you tell a billion-dollar client that their system might be unfair to disabled or neurodivergent candidates — or that it subtly filters out people with regional accents — if they’re the ones keeping your lights on?
“Technically independent. Financially dependent.”
Even the best-intentioned audits exist in a system of incentives. Consultants are still vendors. And vendors, as a rule, don’t bite the hand that signs the cheque.
And let’s be honest — no one wants to upset the wrong billionaire or, heaven forbid, Orange Man. We might end up with a tariff on “algorithmic scrutiny” by morning.
We’ve seen what happens when that kind of dynamic goes unchecked.
Remember the 2008 mortgage crisis? Credit rating agencies handed out AAA ratings to financial garbage because — surprise — the companies selling those toxic assets were the ones paying for the scorecards [7]. The illusion of objectivity collapsed. And so did the economy.
We’re not saying AI hiring will crash global markets. But when people’s livelihoods are being shaped by black-box systems reviewed by paid partners, the parallels are hard to ignore.
“Trust doesn’t come from a contract. It comes from accountability.”
So if you're wondering why these tools keep getting deployed — even with unresolved bias, vague metrics, and suspect audits — maybe it’s not just the companies.
Maybe it’s because the people building them aren’t being told to stop. In fact, they’re being told to go faster.
The Shifting Sands of AI Safety: A National Priority or a Nuisance?
The fight for ethical AI isn’t just happening in company town halls or consulting firm blog posts. It’s playing out at the highest levels of government. And the current message from the top? Move fast, and break… guardrails.
Under the Trump administration, the U.S. has sharply shifted its AI priorities. Gone are the calls for “responsible AI” and fairness. In their place: a race to dominate global AI markets and root out so-called “ideological bias” [8].
The National Institute of Standards and Technology (NIST), the very body tasked with overseeing AI safety, was recently instructed to eliminate references to “AI safety,” “responsibility,” and “fairness” from its partner expectations [9]. Yes — the AI Safety Institute is being told to stop talking about safety.
Instead, they’ve been tasked with building tools to “expand America’s global AI position” [10], following a January 2025 executive order aimed at eliminating anything seen as a “barrier to American AI innovation” [10].
“Safety isn't a barrier to innovation. It’s the foundation for trust.”
Critics warn that stripping away these protections opens the door to unchecked, discriminatory algorithms — including those making hiring decisions [11]. But proponents dismiss the concern. After all, safety audits are “inexpensive,” and what’s a little bias compared to GDP growth? [12]
But when it’s your application being discarded, or your voice being mistranslated by an algorithm, “inexpensive” starts to feel like code for “ignore it.”
This isn’t just a policy change — it’s a worldview shift. One that sees friction as failure and guardrails as bureaucracy. And it sends a clear message to companies building AI hiring tools:
You’re not just allowed to move fast.
You’re being incentivised to.
So when you’re wondering why a platform like HireVue still feels like a digital interrogation booth with no feedback, no accountability, and no second chances — maybe it’s because the people at the top decided that speed wins. Even if people lose.
The Human Cost
I get it. Companies want efficiency. They want scalable hiring. They want to reduce bias. But somewhere in the pursuit of “better,” we stripped away the human from Human Resources.
Real interviews are stressful — but at least they’re a conversation. At least there’s nuance. A moment. A spark. HireVue replaces that with something colder. Something measurable. Something unaccountable.
And if you’re neurodivergent? The challenges are amplified. Time limits, lack of clarification, unclear expectations — it’s a system designed around neurotypical behaviour [13]. If your accent doesn't match the model’s training data, or if your connection stutters, you’re already behind.
The ACLU recently filed a complaint against Intuit and HireVue, alleging their AI hiring technology works worse for deaf and non-white applicants [14]. This complaint highlights concerns that differences in speech patterns, accents, and communication styles can lead to biased outcomes. This isn't just about a "feeling" of unfairness; it's about demonstrable adverse impact on certain demographic groups, echoing concerns that have been raised by numerous studies on algorithmic bias in hiring.
“Efficiency at scale sounds great — until you realise you’ve scaled dehumanisation.”
Why Do We Accept This?
Maybe it’s job scarcity. Maybe it’s tech hype. Maybe we’ve just gotten used to being scanned, scored, and sorted like digital livestock.
Barcoded, processed, and evaluated for “emotional tone” like some sort of psychological meat quality test.
“Smiles: slightly forced. Eye contact: inconsistent. Confidence: medium-rare.”
At some point, we stopped interviewing.
We started optimising — for algorithms, for word clouds, for vibes we can’t see but are somehow being graded on.
And when you start tailoring your voice, posture, and word choice to please a machine — not a person — we lose something important.
We lose the conversation.
We lose the humanity.
We lose the basic dignity of being understood — not just measured, logged, and archived for audit purposes.
“Authenticity is hard when you’re performing for a robot with a spreadsheet.”
Final Thoughts
I’ve talked to walls before — but at least they didn’t reject me via algorithm.
“The system doesn’t need to hate you to hurt you.”
So here’s my message to other job seekers: if you’ve struggled with HireVue, it’s not just you. This system isn’t built to understand you. It’s built to filter. And often, it filters out the very things that make you human.
“We’re not asking for a standing ovation — just a human response.”
We deserve hiring processes that see people, not probabilities. That reward clarity, not conformity. That give you a chance to clarify, connect, recover — not just auto-fail because you blinked wrong in the first five seconds.
We need to talk about this. Loudly. Repeatedly.
With our real voices.
Not just into a camera lens.
And definitely not for a machine that was never listening.
For My Intern Friends: Surviving the Black Box
If you're staring down a HireVue interview and feeling the existential dread of talking to a blinking dot, here are a few tips to help you survive the void — and maybe even score that callback.
1. Treat it like theatre, not a conversation.
You won’t get nods, “mm-hmms,” or follow-ups — so front-load your clarity. Say what you mean, then say it again (briefly) with structure: "Here's the problem, here's how I approached it, here's what happened."
2. Practice with a timer and webcam.
Simulate the awkward silence. I used the Apple timer on my phone — set for 1–2 minutes — then just stared at it while answering questions like I was defusing a bomb. It helped. No audience, no feedback — just me, a screen, and the crushing awareness that somewhere, an algorithm was judging my vocal tone.
3. Smile — but don’t force it.
Yes, it’s weird. But a calm, confident demeanour helps signal “composed and hireable” to whatever behavioural model is watching. Think video cover letter, not hostage tape.
4. Use STAR — even when the question doesn’t ask for it.
That’s Situation, Task, Action, Result. It gives your answer structure, which helps you stay on track — and it helps the algorithm identify story arcs in your response.
5. Review real HireVue questions in advance.
Sites like Coursera 👉 this one have lists of common questions by role. Don’t memorise answers — but do prep your go-to stories for teamwork, problem-solving, leadership, and failure.
6. Learn from the greats (or pretend to).
This YouTube channel taught me everything I needed to know: stare into the void, read a script, and deliver it like you're leading a nation. It's a teleprompter training video for aspiring anchors or presidents — perfect prep for HireVue.
Honestly, after five sessions I felt ready to address the nation, launch a stimulus package, and maybe even explain my gap year with gravitas.
“My fellow stakeholders… in Q4, I overcame adversity by aligning cross-functional synergies under tight deadlines.”
Honestly, if the next U.S. election isn’t run through HireVue, it’s a missed opportunity. At least then we’d have a fair metric for comparing the real candidates to Mark Zuckerberg — At least then we’d all get to see which candidates blink too much under pressure.
“Candidate A: Answered clearly. Blinking rate: suspiciously low.”
Let’s face it: we’re not being judged like humans. We’re being scored like deepfakes trying to pass a CAPTCHA.
“I did not have inappropriate pauses with that question.” — You, to the algorithm, 2025
7. And if all else fails?
Remember: this isn’t the final boss. It’s just one gatekeeper.
You’re more than your sentence flow, your vocal tone, or how long you paused before answering.
You’re not failing an interview.
You’re debugging a system that wasn’t designed for you.
References
[1] Forbes Technology Council. (2025, March 7). The Black Box Problem: Why AI in Recruiting Must Be Transparent and Traceable. Forbes. https://www.forbes.com/councils/forbestechcouncil/2025/03/07/the-black-box-problem-why-ai-in-recruiting-must-be-transparent-and-traceable
[2] Dastin, J. (2018, October 10). Amazon scraps secret AI recruiting tool that showed bias against women. Reuters. https://www.reuters.com/article/us-amazon-com-jobs-automation-idUSKCN1MJ08G
[3] HireVue. (n.d.). Ethical AI. HireVue. https://legal.hirevue.com/product-documentation/ai-ethical-principles
[4] HireVue. (2023, January 17). HireVue to Engage DCI Consulting Group to Audit Algorithms for Bias. HireVue Newsroom. https://www.hirevue.com/press-release/hirevue-leads-industry-in-fair-and-ethical-hiring-practice-engaging-external-auditor-dci-consulting-group-for-external-bias-audit-of-algorithms [5] O'Neil, C. (2021, January 28). ORCAA's Audit of HireVue. ORCAA. https://www.hirevue.com/resources/template/orcaa-report
[6] Landers, R. N. (2021, April 1). Landers Workforce Science LLC Completes Independent IO Audit of HireVue Assessments. https://www.hirevue.com/press-release/independent-audit-affirms-the-scientific-foundation-of-hirevue-assessments https://landers.tech/
[7] Andrews, E. (2009, February 16). How the Credit Rating Agencies Contributed to the Financial Crisis. The New York Times. https://truthout.org/articles/the-indisputable-role-of-credit-ratings-agencies-in-the-2008-collapse-and-why-nothing-has-changed/
[8] Knight, W. (2025). Under Trump, AI Scientists Are Told to Remove ‘Ideological Bias’ From Powerful Models. Available at: https://www.wired.com/story/ai-safety-institute-new-directive-america-first/ (Accessed: 26 May 2025).
[9] The White House. (2025, January 23). Executive Order: Removing Barriers to American Leadership in Artificial Intelligence. https://www.whitehouse.gov/presidential-actions/2025/01/removing-barriers-to-american-leadership-in-artificial-intelligence/
[10] Clifford Chance. (2024, November 7). AI Pulse Check: Will the Biden Executive Order on AI Survive the Trump-Vance Administration? https://www.cliffordchance.com/insights/resources/blogs/talking-tech/en/articles/2024/11/will-the-biden-executive-order-on-ai-survive-the-trump-vance-administration.html
[11] See reference [14] (ACLU complaint) as a key example of potential harm.
[12] The White House. (2025, January 23). Executive Order: Removing Barriers to American Leadership in Artificial Intelligence. https://www.whitehouse.gov/presidential-actions/2025/01/removing-barriers-to-american-leadership-in-artificial-intelligence/
[13] Glouberman, M. & Goldberg, M. (2024, June 4). How to Make Job Interviews More Accessible. Harvard Business Review. https://hbr.org/2024/06/how-to-make-job-interviews-more-accessible
[14] ACLU. (2024, May 14). Complaint Filed Against Intuit and HireVue Over Biased AI Hiring Technology That Works Worse For Deaf and Non-White Applicants. https://www.aclu.org/press-releases/complaint-filed-against-intuit-and-hirevue-over-biased-ai-hiring-technology-that-works-worse-for-deaf-and-non-white-applicants
