Once upon a time, the labor market was mostly local. Businesses placed small ads in newspapers, received a handful of applications, and made their selections from there. Each stage of the process, from the cost of advertising to writing and mailing applications, to having personnel review them, involved significant marginal costs. There was friction—job seekers had to spend money to send applications, or for some jobs they had to show up in person to apply. The system wasn’t scalable from either side.
From today’s perspective, the job market of yesteryear may seem like a lost paradise. With these costs essentially eliminated, HR departments now receive hundreds of applications for every position. Applicants can easily submit their materials online to companies across the country or even across continents. They can click “apply” on LinkedIn or Indeed without even reading the ad (and many do).
Remote interviews removed the need for physical presence. Automated rejection systems emerged in the 2010s to handle the flood of digital applications by scanning for keywords and surface-level professionalism, such as avoiding typos in cover letters and CVs. Now, with AI, anyone can generate polished application materials filled with the right keywords, tailored to match job descriptions and pass through these filters. The result is flawless-looking but often soulless documents that no human would be impressed by, yet the automated system pushes them forward to the interview stage, often without any human review at all.
So far, companies haven’t truly adapted. They can’t ditch automated systems, as that would bury staff under an avalanche of AI-generated slop. But they also can’t rely solely on them, since filters increasingly weed out authentic, personally written applications in favor of generic LinkedIn-style content optimized by AI.
The obvious next step, using AI to filter applicants more intelligently, is currently stymied by regulation. In the EU, the AI Act has preemptively classified AI-driven hiring systems as “high risk,” making them virtually unusable due to excessive compliance requirements. The US isn’t faring much better. Because large language models are pattern-recognition machines, using them in hiring decisions risks violating anti-discrimination laws that prohibit filtering for certain patterns. Combining AI filtering with current legal standards opens companies to endless lawsuits, so even those desperate for such systems shy away from them.
The result is a broken system. Applicants are trapped in a demoralizing game, sending hundreds of applications with little chance of standing out. Meanwhile, companies struggle to identify genuinely strong candidates in a sea of indistinguishable, AI-crafted submissions.
There may be two potential ways forward.
First, companies could begin charging a small fee for each application, perhaps funneled to charity to avoid the appearance of monetizing recruitment. This would introduce a layer of self-selection, encouraging only serious applicants to apply. Those who do would be more likely to invest real effort in their applications, and companies could, in return, promise a personal review.
Second, the AI labor market might evolve into a functioning system, not as an internal selection tool within companies but as a broker platform that matches candidates with employers. This could sidestep anti-discrimination laws, since the platform would facilitate connections rather than make final hiring decisions. Ideally, such a system would allow people to present themselves authentically, rather than conforming to cringe-worthy corporate stereotypes, and AI would help match them with roles that truly suit them. But getting there will require substantial advances in AI and a wave of entrepreneurial innovation.
Still, one can hope: a future where talented individuals are matched with meaningful jobs without enduring the increasingly absurd game of applications.
