The other day I saw a post about using LLMs for candidate screenings, i.e. fully unattended systems, no human-in-the-loop. As in many current discussions about AI, opinions seemed to split into for and against. As someone who has researched and written about human-centered AI usage, I thought this was a good opportunity to explore issues specific to AI in hiring. I did, after all, work a stint at Indeed.
Let me be clear. I am not against using LLMs in hiring. I’ve been a technology consultant for *checks notes* a million years. I have witnessed promising new technologies applied in ways that ultimately limited their potential to serve us and improve our lives. The goal is smart, human-focused application. This means moving beyond the hype and fear to identify sustainable, win-win use cases that serve both organizations and candidates.
The current trend of deploying unattended LLMs for first-tier screenings is a strategic misstep. Even in an employer-favorable market, it risks brand damage and candidate alienation. When the market inevitably swings back, this practice will become active poison for talent acquisition. A 2025 Harris Poll shows 84% of candidates prefer a human for the initial screen, and Pew Research found 66% would not apply to a company using AI in hiring decisions. This isn’t sustainable innovation; it is a short-term gambit with long-term consequences. No bueno.
It is all well and good to critique what is broken. Real value, however, lies in envisioning what comes next. So let’s ask the question: what might a more human-centered AI collaboration in hiring actually look like?
The shift required is fundamental. We must stop asking how AI can replace humans and start designing how it can augment them. Consider the common frustrations on both sides of the interview table. A candidate often struggles to decode the true priorities hidden within a job description. A hiring manager, juggling endless meetings, may have only glanced at the résumé before the call. This creates an inefficient dance where both parties waste time simply establishing basic alignment.
This is where a transparent AI copilot shows promise. Imagine a system that analyzes the résumé against the job description before the interview, surfacing qualitative matches (not just keywords) and potential gaps. That analysis is then shared on screen with both the candidate and the hiring manager at the start of the conversation. The LLM provides suggested talking points or areas for clarification that the hiring manager can use or overrule. It becomes a collaborative tool. The hiring manager gets cognitive relief to lead a more focused discussion. The candidate gains agency to clarify or expand on points in real-time. The AI isn’t the judge; it’s the moderator ensuring both parties start on the same page.
Something like this was once logistically challenging, but post-COVID, the ubiquitous nature of remote meetings makes an AI-assisted first-tier interview a viable implementation. If successful, the candidate can progress to more open-ended rounds in person or remotely. I don’t propose this as a baked solution, but as a thought experiment for how human-centered AI might be operationally leveraged. No doubt some companies are already thinking this way or even building it. The core principle is that sustainable application requires transparency and shared benefit for both the hiring manager and the candidate.
The value here is derived from two non-negotiable principles: radical transparency and shared benefit. The hiring manager gains efficiency through reduced cognitive load and sharper focus. The candidate gains agency through visibility into the process and the opportunity to clarify context. This isn’t about replacing human judgment; it’s about enhancing its quality and fairness. This approach is best suited for roles requiring nuanced evaluation, not high-volume, low-complexity hiring where its advantages are diminished. No worries, we can come up with empathetic alternatives for that as well…with a little nuanced thought.
Humans are true experts at that.
The debate around AI in hiring does not need to be a battle between evangelists and skeptics. The most strategic path forward rejects this false dichotomy. We do not need to be uncritical advocates who deploy technology blindly, nor do we need to be reactionary opponents who dismiss its potential entirely. The goal is to thread the needle to enhance efficiency without eroding trust. Good vibes only.
Final thought from my heart to those struggling in this job market who might be reading this. If you are unemployed, or underemployed, we’re in an objectively soft employment market and AI is now crashing the party. The odds are stacked, but there are many of us in tech working on this. We see you. Take a deep breath. Remember who you are; the same amazing person you were before the market shifted. You are unique, powerful, and you’ve got this. The following song is my encouragement to you. Push on. ??