AI is changing both sides of the job hunt. Applying more won’t fix it

A few months ago, a candidate wrote to me on LinkedIn. Eleven years of experience, a genuinely strong resume, over 200 applications sent in four months, three interviews to show for it. She asked me what she was doing wrong. I get some version of this message every week, and my honest answer is usually the same: nothing. She wasn't doing anything wrong. She was playing a game whose rules had already changed, using a strategy built for a hiring process that stopped working years ago.

I've spent 11 years in India's career guidance space through Mentoria, working with over 3.5 lakh professionals. More recently, I've also been building AI systems on the recruiter's side of hiring, which means I've watched what actually happens to an application after someone hits submit. That dual view is why I no longer trust most of the job search advice still being handed out.

India’s graduate unemployment problem is bigger than skills alone

Graduate unemployment is often framed primarily as a skills problem. But the data suggests the challenge is more complicated.  According to a report from Azim Premji University, graduate unemployment among 15 to 25 year olds sits close to 40%, and around 20% among 25 to 29 year olds. Only about 7% of graduates who report themselves unemployed land a permanent salaried job within a year. Between 2004 and 2023, India added roughly 5 million graduates a year, but only around 2.8 million of them found work in that same stretch.

Roles exist. Part of the problem is the process that's supposed to connect a qualified person to the recruiter hiring for that role, and that process is now being rebuilt by AI on both sides at once.

Do ATS systems really reject 75% of resumes?

No, and this is the myth I correct most often when I'm talking to job seekers directly. The 75% figure traces back to a marketing claim from a company that made it in 2012 and shut down the following year. The widely reported figure does not appear to be backed by credible published research. A 2025 study of 25 recruiters across major ATS platforms found that 92% don't configure their systems to auto-reject resumes over formatting or missing keywords.

What I've actually seen building on the recruiter side is more mundane, and honestly harder to fix with a resume template. A  study found that 88% of employers believe their own hiring technology filters out qualified people simply because those candidates didn't use the exact search terms a recruiter typed in. I've sat with recruiting teams running searches against more than 500 applications for a single role. Nobody's building this to be unfair.

They're running a keyword search because keyword search is the simplest method that scales to that volume, and at that volume, good people get missed. I saw it happen to a candidate we'd guided through three separate job changes: someone with exactly the right experience, filtered out because her resume said "customer success" and the search ran on "client success."

Why doesn't applying to more jobs improve your odds?

Career advice has spent years telling job seekers to widen the net: apply to more roles, increase the number of shots on goal. I understand the instinct. It feels like effort, and effort feels safer than the alternative, which is admitting the system itself is the problem. But a generic application sent to fifty roles performs worse in a keyword search than a tailored one sent to five, because the tailored version actually contains the language a recruiter's system is searching for. I watched this play out with the candidate I mentioned earlier. Two hundred applications, almost none tailored, three interviews. Volume was never her problem. Precision was.

Does AI make the application volume problem better or worse?

This is where the conversation gets more complicated than most people are willing to sit with. AI now makes it trivially easy to generate and fire off far more applications than any person could manually produce in a day. If volume was already a losing strategy against a manual keyword filter, AI-accelerated volume doesn't solve that. It scales the same mistake.

A recruiter's inbox doesn't get more discerning because five hundred technically submitted, poorly-matched applications arrived instead of fifty.

The variable that actually matters here isn't whether the process is manual or automated. It's whether the tool is optimising for precision or for volume.

What happens when recruiters start using AI too?

More interesting is what happens when recruiters start using AI too.

It's the question I think about most in my own work. Candidates aren't the only ones adopting AI. Recruiters and hiring platforms are increasingly using AI for resume screening, ranking, and early-stage outreach too. So now you have two automated systems, one on each side of the same transaction, and neither was built with the other in mind.

I've seen what that looks like from the recruiter-facing side: an AI-drafted application meeting an AI-driven filter, with no human judgment touching the interaction at any point, unless someone deliberately designs the systems to prioritise real fit over surface-level keyword matching. That's a genuinely different problem than "how do I beat the filter," which is the framing most career content is still stuck on. The real question is what happens to hiring when both sides hand the first pass to a machine, and the two machines aren't optimising for the same outcome.

So what should job seekers actually do?

I don't think the answer is to reject AI on either side of this. The old process was already too high-volume and too poorly matched for a purely manual approach to fix, long before AI entered the picture. But an automated arms race that just produces more mismatched applications, faster, on both ends, isn't progress either. It simply scales the existing mismatch.

The candidates who do well over the next couple of years won't be the ones submitting the most applications: human-written or AI-generated. They'll be the ones using AI selectively to improve fit: better-matched applications, more direct access to the person actually hiring, and more of their own time spent on the things that still need a human, such as interview prep, negotiation, and deciding which offer is right. That candidate I mentioned earlier switched her approach about six weeks ago. Fewer applications, each one built around a role she actually matched. She had two offers within a month.

Manual job hunting is struggling because it was built for a hiring process that no longer exists. The fix isn't more automation on the same broken metric. It's tools on both sides of the table finally optimising for fit instead of speed.

Nikhar Arora is Co-Founder of Arya by Mentoria