A recruiter posts a new role on Monday. By Wednesday there are 200 applications in the inbox. By Friday its past 400.
For about five minutes, that feels great. More candidates should mean more choice and a faster hire.
Then three weeks go by and the role is still open. The recruiter has read hundreds of near-identical resumes, booked a few calls nobody showed up for, and ended up more tired than when this started.
People are starting to call this the AI doom loop, and once you see it, it’s hard to unsee: AI helps candidates apply to everything, AI helps recruiters filter everything, and somehow nobody’s actually getting hired.
So what’s actually going on here?
Really, it’s two AI tools pulling in opposite directions at the same time.
On the candidate side, people are using AI to apply to dozens of jobs in the time it used to take to write one cover letter. Paste in a job link, and a tool rewrites the resume, writes a new cover letter, and hits “Easy Apply.” The whole thing takes under a minute.
On the recruiter side, most teams now lean on AI inside their applicant tracking system just to survive the volume. Keyword filters. Auto-ranking. Chatbots that ask the first round of screening questions before a human even looks at a resume.
Both sides are just trying to save time. But when both automate at once, both sides start writing for the algorithm instead of the actual job. Candidates learn which keywords the ATS is scanning for and stuff their resumes with them. Recruiters lean even harder on filters because there’s no other way to get through 400 applications by Friday. And the thing that actually matters, whether someone can do the job, gets lost in the middle.
Why More Applications Don’t Mean Better Candidates
It gets worse from there. When candidates notice their response rate drop, they don’t apply less, they apply to more jobs and lean on AI even harder to do it. Recruiters notice how similar every application starts to look, so they add another filter, or bring back an extra interview round just to be safe. Each side is reacting to the other, and neither fix actually solves anything. That’s the loop part.
The numbers back this up. Hiring benchmarks from earlier this year put the average job opening at 300+ applications, roughly triple what it was just a few years back. Interview rates have fallen by about half in that same stretch, and it now takes eight to ten weeks to fill roles that used to move much faster. More people are applying. Somehow, fewer of them are getting hired.
This isn’t just a big tech problem
It’s easy to assume this is a big-tech problem, but mid-size companies and staffing desks feel it just as hard, often worse, without ten recruiters to throw at the pile. Here’s what it actually looks like week to week:
- Time-to-fill keeps going up instead of down, since sorting the noise now takes longer than finding candidates used to.
- Genuinely good candidates get auto-rejected because their resume didn’t use the exact keyword the system was scanning for.
- Candidates ghost interviews more often, partly because they applied to so many roles they don’t remember which company this even is.
- Recruiters are burning out from reading hundreds of resumes that all sound the same, for very little payoff.
- When rejections are generic and companies go quiet for weeks, candidates talk about it publicly, and that follows the employer into the next hiring cycle too.
Four stages, one loop: more applications keep coming in, and the seat stays empty.
Why more AI doesn’t fix this
The obvious next move is to add another layer of automation. A smarter filter. A stricter keyword match. Maybe a bot that runs the first interview. But that just speeds up the same loop. It doesn’t break it. Candidates figure out the new filter within a few weeks, applications start looking “qualified” to the algorithm again, and everyone’s back where they started.
What actually helps is using AI for the part it’s good at, sorting and scheduling, and keeping people in charge of the part it isn’t: deciding who’s actually right for the job.
How do you actually break the loop?
A few things tend to separate teams who’ve climbed out of this from teams still stuck in it:
- Look at skills before keywords. A short, structured assessment tells you more about someone in twenty minutes than a resume scan tells you in twenty seconds.
- Build a pipeline instead of leaving the floodgates open. If candidates are already screened before they land in your inbox, there’s a lot less noise to sort through.
- Keep a person at the top of the funnel. AI can rank and sort all day, but someone still has to decide who’s worth an actual conversation.
- Write clearer job posts. Vague requirements pull in vague, mass-applied resumes. Specific ones don’t.
This is basically the idea behind NAM Info’s ADEPT framework. Every consultant goes through a technical assessment, a business-fit check, and background verification before a client ever sees their profile. So instead of sorting through a pile of 400, you’re choosing from a shortlist that’s already been vetted.
Where does this leave recruiters?
More applications was supposed to make hiring easier. Right now, for a lot of teams, it’s doing the opposite. Both sides keep automating, the noise keeps piling up, and the role stays open longer than it should.
The point isn’t to use less AI. It’s to make sure a person is still the one deciding who’s worth a conversation, while AI handles everything else around that.
If your team is drowning in applications and still can’t close a role, a smarter filter probably won’t fix it. A better pipeline will.
Quick questions people ask about this
What is the AI doom loop?
It’s what happens when AI-assisted job applications and AI-powered resume screening run into each other. Recruiters end up with far more applications than before, but sorting through them takes so long that roles actually take longer to fill, not less.
Why are recruiters suddenly getting hundreds of applications per role?
AI tools let candidates apply to dozens of jobs in minutes with auto-generated resumes and cover letters. Postings that used to get 50 to 100 applications now regularly pull in 300 or more.
How can a company actually fix this?
Not by adding more filters. Teams that handle this well use AI for scheduling and sorting, keep a person making the actual hiring calls, and pull from talent pools that are already screened instead of sorting through everything that comes in cold.
