Ask most people what AI has done to hiring and they’ll tell you it made it faster and cheaper. More candidates, less effort, better matching. It’s a tidy story. It’s also wrong — or at least, it’s looking in the wrong place.
AI has changed hiring profoundly. But it didn’t lower the cost of getting a hire right. It moved that cost somewhere the finance team isn’t looking, and in a lot of cases it made it bigger. The firms that understand this are quietly pulling ahead. The ones still selling "we use AI to source faster" are automating the least valuable part of their own job.
Here’s what’s actually happening.
The resume stopped being a signal
A competitive role today can draw more than a thousand applications in a matter of days. That isn’t a sign of a healthy market — it’s a sign that applying has become nearly free. Roughly seven in ten job seekers now use generative AI somewhere in their search, and a large share use it to write or optimize the resume itself.
The result is a stack of applications that are all polished, all keyword-perfect, and almost impossible to tell apart.
The people on the receiving end feel it. Around nine in ten hiring managers say AI has increased their workload, not reduced it, and most say screening now takes longer than it used to. More than three-quarters say AI has made it harder to judge whether a candidate is even authentic.
Read that again: the tool that was supposed to speed up screening has made the single most important job in screening — telling real from packaged — measurably harder.
So the "cheap" channels aren’t cheap. The cost didn’t disappear; it turned into hours. Hours spent sorting noise, chasing false positives, and — the expensive part — occasionally missing a genuinely strong candidate because they were buried under two hundred people who wrote a better prompt.
The fakes got good
There’s a newer cost that didn’t exist a few years ago: making sure the person is real.
Deepfake candidates and AI-assisted interview fraud moved from novelty to genuine problem in 2026 — fabricated identities, stand-in interviewees, and in some cases AI running quietly on the other side of a video call. It’s concentrated in remote and technical hiring, which is precisely where a lot of engineering and technology roles now sit.
Verification — simply confirming that the competent person you interviewed is the person who shows up — is now a line item on the true cost of a hire.
The rules are arriving on a clock
And then there’s compliance. The European Union’s AI Act brings hiring AI — CV filtering, candidate ranking, interview scoring — under its high-risk regime as of August 2, 2026, with real obligations: human oversight, bias testing on your own data, documentation, vendor due diligence, and penalties that reach into the millions.
It applies to companies outside the EU that hire into it. New York City already audits hiring algorithms. Wherever your business touches, the regulatory direction is one-way, and "the software vendor handles it" is not going to be an answer.
None of this is an argument against AI. We use it, and so should you. It’s an argument against the assumption that AI took the cost out of hiring. It didn’t. It moved it, and it added risk on top.
The part that should worry a leader most
Add these up and you get something more serious than wasted hours. You get a higher chance of hiring the wrong person.
We’ve written before about how the slow hire becomes the bad hire — how a role left open too long turns into pressure, and pressure is what makes good leaders settle. AI pours fuel on exactly that fire.
When your funnel is full of applications engineered to pass the screen, when a resume no longer tells you much, and when you’re not entirely sure the person is who they say they are, the odds of a confident, expensive mistake go up.
A bad senior hire is already the most expensive, hardest-to-undo decision a leader makes. An AI-flooded process makes it more likely, not less.
If you want to see what that costs in dollars, we built two calculators for it — the cost of a vacancy and the cost of a bad hire. The numbers tend to end the debate.
What actually holds up
Here’s the part almost no one in our industry is saying out loud: the things AI is automating in recruitment were never where the value was.
Sourcing, keyword matching, forwarding resumes, booking calls — that was always the commodity layer, and AI is right to eat it. What AI cannot do is the thing hiring actually turns on: reading a person. Judging whether someone’s competence is real and deep or polished and thin.
Understanding whether a leader can carry a function through two years of change. Knowing a market well enough to tell a client not just who’s available, but who’s good and why. That work is human, and as AI floods everything around it, it becomes more valuable and more scarce — not less.
That’s why the resume-forwarding agency is genuinely threatened by AI, and a firm built on judgement is not. It’s also why our answer to "how do you deal with AI" isn’t a piece of software. It’s a live, unscripted conversation with someone who knows your field — the one setting AI can’t pass by writing a better prompt. We use AI to cut the noise. We use judgement to make the call.
Testing for the thing clients now actually ask about
There’s a flip side to all of this, and our clients raised it before we did: they don’t just want candidates who aren’t faking with AI — they want candidates who can genuinely use it. "Is this person actually AI-fluent, or do they just say so?" is now a real hiring criterion, especially in engineering and technology.
So we assess for it directly. As part of how we vet, we look at whether a candidate can use AI well in their actual work — not whether they’ve touched a chatbot, but whether they can tell when the output is wrong and verify it, whether they’ve genuinely changed how they work, whether they know where not to use it, and, for leaders, whether they can set the direction for a whole team.
And because the same conversation that reveals genuine fluency also reveals faked competence, we get both answers at once: is the skill real, and is the person.
That’s the quiet advantage of a model built on talking to people properly. In a market where a resume can be written by a machine and an interview can be sat by one, the firm that reads people is holding the one thing that still works.
PHM Search is an executive and specialist recruitment firm for mid-market engineering, construction, and technology companies across Canada and the USA. Since 2000, we’ve helped leaders hire the people their plans depend on — and avoid the ones that look right on paper. If you’re hiring into a role where getting it wrong is expensive, start a conversation, or see how our executive search is built to remove the pressure that manufactures bad hires.
