She had done everything right.
Ten years of excellent product work. Name brands on the resume, the kind that used to get you a callback before the recruiter finished their coffee. She was laid off eight months ago in a round that hit half her department, moved to New York, and started over.
Now she can't get a phone screen.
"In the past," she told me, "I didn't need to do anything except signal that I was interested. I'd get a call the same week." Today, she's applying into silence. Not because her work got worse. Because the game changed underneath her while she was busy doing the work.
This is the AI crunch. And if you've been paying attention, you've felt it, even if nobody's said it plainly.
The Signal Collapsed
For decades, the job application was an imperfect but functional filter. Writing a good cover letter took effort. Effort was a proxy for care. A referral put someone's reputation on the line. These weren't perfect signals, but they were costly to produce, which meant they were worth judging.
AI made faking it free.
Now a candidate can generate a tailored resume, a personalized cover letter, and interview-ready answers for fifty different roles before breakfast. Presentation quality, the thing that used to tell you something about how someone thinks, tells you almost nothing now. Applications per posting have more than doubled in five years.* Recruiters who were already stretched are now buried.
Somewhere in that pile, underneath the bots, the bulk-appliers, the AI-polished fakers, is the person who can actually do the job.
Good luck finding them.
HR's Answer Made It Worse
The industry's response to AI flooding the top of the funnel was to deploy more AI to filter it. And that makes sense. In the business world we are judged on efficiency metrics. How much cleaner can we make the process? Time saved is money saved for the business.
And so entered the resume screeners, automated scoring, AI-conducted first-round interviews. It was AI-generated applications now being evaluated by AI screening tools.
We were so busy patting ourselves on the back for our efficiency wins that it took a moment to notice that everyone was losing.
We created an arms race and the people caught in the middle, the actual builders, the real operators, are losing the most.
So what did we do? Candidates optimized harder. Filters got tighter. More noise enters the system. The metrics look clean, applications processed, time-to-hire reduced, costs down. What doesn't show up on the dashboard is the perfect person who got screened out because she described her experience in plain English instead of keyword-engineered prose.
The hiring process has become a test of one very specific skill: the ability to pass the hiring process.
For almost everything else, it's a measurement error dressed up as rigor. And by adding more filters to a signal problem you just get faster at making the same mistake.
What Actually Got Lost
Strip away the technology for a second and ask what this was always supposed to do.
Find people who can do excellent work.
Not people who can describe excellent work. Not people who perform well in structured interviews. People who have done it, shipped the product, rebuilt the broken system, made the call that saved the quarter, and can do it again.
The evidence to back up that work exists, but it is trapped in people's memories. There is no current infrastructure to capture it and make it portable.
The people doing the most important work, the quiet, compounding, judgment-heavy work that AI still cannot replicate, are invisible. And the people who have mastered the performance of competence are everywhere.
And that is the job desert we are all so frustratingly experiencing. Tons of applicants and zero signal.
The Answer Isn't More AI
Every solution being floated right now reaches for the same tool that caused the problem (and keeps making it worse). Now we have people reading AI answers live in interviews. Or worse, the recruiter asking the questions is AI. It's absurd enough to make you want to scream.
You cannot AI your way out of an AI problem. Not this one.
This isn't meant to be an anti-AI take. In this scenario what's broken is how we show proof. The only thing that holds up when everything else can be faked is a real human being, with their own reputation on the line, saying: I was there. I saw what they did. It was real.
Not a skill badge. Not a keyword match. A person, vouching specifically, for work they witnessed firsthand.
That signal exists. For most people it is just scattered, inaccessible, and slowly disappearing in a landscape where you need it and stat.
Careers need a new kind of infrastructure entirely, one built around verified human proof. Where the work you actually did, witnessed by people who were actually there, becomes something portable. Something durable. Something yours.
That infrastructure does not exist yet, but it needs to.
*Business Insider / Novorésumé, February 2026; Glassdoor Hiring Data