The numbers don’t lie. Since the mainstream arrival of generative AI, hiring teams at mid-to-large companies have reported anywhere from a 40% to 300% spike in application volume for a single role. And it’s becoming common practice to find most of those resumes are written, at least in part, by an AI tool like CoPilot or Claude.
So now we’re in a strange new arms race. Candidates use AI to write their resumes. Companies use AI to screen them. And somewhere in the middle, great talent is getting filtered out. Not because they weren’t qualified, but because they didn’t know the right prompt.
This isn’t a rant against AI-generated resumes. And, actually, I feel confident this will work itself out as agentic AI grows in use. But, for now, hiring teams need a strategy guide so they can stop drowning in applications and start making smarter decisions, without accidentally building a system that rewards AI literacy over actual job performance.
What You’re Actually Dealing With
Not all AI-assisted resumes are equal, and treating them that way is a mistake.
There’s a meaningful difference between a candidate who ran a rough draft through Grammarly, one who used AI to help structure a career transition narrative, and one who fed a job description into ChatGPT and copy-pasted the output wholesale. Your screening strategy should account for that spectrum because the first two candidates might be exactly who you’re looking for.
The real problem isn’t AI. The real problem is that a flood of polished, keyword-optimized, structurally identical documents has made it nearly impossible to identify the best candidates. Your ATS is drowning. Recruiters are numb. And there is a real frustration among hiring managers because so many of the candidates with great resumes crash during the first conversation.
Using AI Smarter To Screen Applicants
Take these steps to get past all the noise and to the best candidates:
1. Stop screening for polish. Start screening for specificity.
AI excels at producing generic excellence. It writes confident bullet points, uses industry keywords fluently, and navigates ATS filters with surgical precision. What AI struggles with (unless prompted very carefully) is specificity.
Configure your AI screening tools not to reward keyword density, but to flag and surface specific, quantified, contextual claims. A bullet that says “Led cross-functional team to deliver $2.3M digital transformation initiative 6 weeks ahead of schedule at a 47-person regional bank” is almost certainly human, even if AI-polished. A bullet that says “Demonstrated strong leadership in driving organizational change initiatives” is almost certainly AI-generated filler. Train your screening prompts accordingly.
2. Add a structured, AI-resistant intake question.
Before a resume ever hits a human’s inbox, add a single required short-answer question to your application. Not “Why do you want to work here?” That’s an AI layup. Instead, ask something that requires real experience to answer well:
“Describe a specific moment in your last role when something you were responsible for didn’t go as planned. What happened, and what did you do?”
Or: “What’s something about how [your industry/function] works that most people outside of it get wrong?”
No question is really AI-proof, but you can find those that are “AI-resistant”. A candidate who can answer them with genuine texture and specificity has at least engaged with the question as a human. Use AI to rapidly scan these responses for the specificity signals described above and locate the ones worth reading.
3. Use AI to build better shortlists, not final decisions.
There will be a time when your AI agent identifies prospects, screens them, and recommends those to hire (FYI, the technology exists and that time is now). For those not using that technology, AI is the tool to shorten the list of candidates to review.
Implement AI screening to get from 800 applications to 40 or 50, not to get from 800 to 8. The goal is to eliminate the clear mismatches and surface a manageable pool for human review. The moment you let an algorithm make the call on who’s worth a conversation, you’ve introduced a bias layer you probably can’t audit and almost certainly can’t defend. Note: I know some might say the human is the bias and the AI will eliminate it. But, right now, convention and proven process reigns.
Think of AI as your first-pass analyst. It’s fast, it’s tireless, and it can pattern-match at scale. But it doesn’t know that the candidate with the non-linear career path who started a failed company is exactly the kind of resilient worker your team needs right now. That’s a human call.
4. Calibrate your AI to your actual top performers — not your ideal candidate profile.
Most AI screening tools are trained on general hiring data or on a job description you wrote. Neither of those is as valuable as this: take your five best hires from the last three years, pull their original resumes, and analyze what those resumes actually looked like before you knew they were great.
Were they traditionally formatted? Were they unconventional? Did they include gaps or lateral moves? Did they come from adjacent industries? Are there any trends you see from this evidence to help you build AI screening criteria? This is worth more than what you pull from an idealized template. You might be surprised by what your best performers have in common that isn’t listed on the job description.
5. Phone screens are a calibration tool, not a courtesy.
There is a growing trend, in the rush to automate, to eliminate screens early in the process. That is a mistake. A brief, structured conversation is still the most efficient way to validate whether a candidate’s experience matches their resume. I’ll save how to determine the person on your video interview is actually real for another time!
Use AI to handle scheduling, note-taking, and post-call summaries. Let it do the administrative work. But give thought to and know the proper points to keep a human voice on the phone. The way someone answers “Walk me through what you actually did day-to-day in that role” tells you more than three rounds of AI screening ever could. Recruiters want to get to know candidates. Provide touch points for them to do this but leave the paperwork to AI. I wrote in detail about this in a recent article, “The Rise of Agentic AI“.
The Talent You’re Most Likely Losing Right Now
Here’s who falls through the cracks of aggressive AI screening, and why you should care:
Career Transitioners. They’ve applied to roles adjacent to their experience, and their resumes don’t match your keyword model, even though their skills transfer directly. AI screening often penalizes them for what’s missing rather than recognizing what’s there.
Senior Professionals. Many experienced candidates have shorter, denser resumes that don’t “optimize” well. They’re not performing for an ATS. They’re assuming their track record speaks for itself. Your AI might be systematically deprioritizing your most experienced applicants.
Candidates from Underrepresented Backgrounds. AI screening tools trained on historical data replicate historical biases. If your top performers have tended to come from certain schools, companies, or career paths, your AI will reward candidates who match that pattern, which narrows your talent pool in ways that compound over time.
AI didn’t create the resume screening problem. Volume did. AI-generated resumes are a symptom of a broken dynamic, not the cause of it. The answer isn’t to build higher walls. It’s to get smarter about what you’re actually looking for and use the right tools at the right stages to find it.
Use AI to work faster. Use humans to work better. And build your screening process around the question that actually matters: not “Does this resume look right?” but “Is there a real person here who can do this job?”
That’s a question worth answering carefully, no matter who wrote the resume.
What’s your team doing to adapt your screening process in the AI era? I’d love to hear what’s working — and what isn’t. Drop a comment below.
ES Talent Solutions helps organizations navigate the intersection of recruiting strategy and emerging technology. Want to discuss how agentic AI could transform your talent acquisition function? Contact Eddie Stewart at estewart@ESTalentSolutions.com. I’m always happy to talk with fellow leaders about building recruiting functions ready for the future.





0 Comments
Onboarding Is the Final Step of Sourcing
Nearly half of all first-year employee departures happen within the first 90 days—and the number one reason isn’t bad orientation or delayed IT setup. It’s the gap between what candidates were promised during the hiring process and the actual day-to-day reality of the role. When recruiting dashboards stop counting at offer acceptance, organizations measure the sale while ignoring the returns. In his latest article, TA Leader and Recruiting Architect Eddie Stewart breaks down why onboarding is the ultimate audit of your sourcing strategy, offering six actionable ways Talent Acquisition teams can bridge the handoff gap, measure 90-day retention, and stop refilling avoidable reqs.
The Demographic Cliff
The current soft job market is a deceptive illusion masking a permanent, structural labor shortage. Driven by decades of declining birth rates, an aging workforce, and an impending “enrollment cliff,” the U.S. labor force is set to shrink by nearly 6 million workers between 2025 and 2032. While AI-generated spam application floods give employers a false sense of candidate abundance, critical sectors like healthcare and skilled trades face structural deficits that technology cannot fix. Stewart urges talent leaders to stop waiting out market cycles and instead adopt proactive workforce strategies—including internal mobility, phased retirements, broader campus outreach, and alternative talent pools.
The Regulatory Roadmap, Part 2
January 1 brings new AI hiring rules in Colorado and California, and they all start with one question: which tools in your funnel score, rank, or filter candidates? Most compliance problems in 2027 will start as inventory problems in 2026.