Imagine a hiring manager looking at a screen on a Monday morning in a glass-walled office in Chicago, London, or Singapore. For a mid-level marketing position, she has 412 applications pending. The first one is opened by her. Next, the second.
By the fifteenth, a strange, creeping sameness begins to set in, rather than fatigue per se. The language is refined. The formatting is neat. All of the phrases are expertly chosen. Nevertheless, none of these documents seem to have been written by a real person.
| Subject | Applicant Tracking Systems (ATS) & AI in Hiring |
| Sector | Human Resources, Recruitment Technology, Enterprise Software |
| Key statistic | Over 98% of Fortune 500 companies use ATS software in their hiring process (Jobscan) |
| Volume | ~11,000 job applications processed per minute on LinkedIn alone |
| Fraud projection | By 2028, an estimated 1 in 4 job applicants may be fraudulent, leveraging AI tools |
| Recruiter ATS adoption | 75% of recruiters use some form of ATS or recruiting system (Capterra) |
| Bad hire cost | 74% of employers admit suffering costs from a bad hire linked to poor skills alignment |
| Reference | Jobscan — Applicant Tracking Systems Guide ↗ |
This is the current hiring environment. It didn’t come all at once. It began slowly over a number of years, picked up speed by the pandemic’s surge in remote work, accelerated by the proliferation of AI writing tools, and now settled into what appears to be a collapse.
Millions of people have worked on the neat, two-page resume at kitchen tables and library desks, but it has been methodically diminished. The peculiar irony is that the majority of the harm is being caused by the very technology designed to improve hiring efficiency.
It’s hard to dispute the numbers. On LinkedIn alone, about 11,000 job applications are processed every minute. More than 98% of Fortune 500 companies use Applicant Tracking Systems, which are programs designed to score resumes against job descriptions, parse resumes for keywords, and determine in a matter of seconds whether a human will ever see them.
These tools are used by 75% of recruiters in the general market. This implies that people do not read the majority of resumes nowadays. They are evaluated by machines. Additionally, the majority of candidates who discovered this have resumed gaming the machines.
As a result, there is an arms race that no one is winning. When candidates feed their experience into AI writing tools, they receive glossy, keyword-rich documents that perform well on algorithms but say very little that is genuine. After that, hiring teams run those documents through ATS software, which was never intended to recognize when a language model has taken the place of a human in the drafting chair.
Because they described their work in straightforward, honest language that didn’t exactly match the exact phrasing the system was trained to recognize, brilliant people with truly relevant experience are filtered out. Additionally, mediocre applicants who possess superior prompt-engineering abilities advance to the interview stage.
The majority of those working in the recruiting sector seem to already be aware of this. At conferences, they discuss it. In Slack channels, they joke nervously about it. However, in the public eye, the resume maintains its position because dismantling it would necessitate rebuilding something, which is difficult, costly, and uncertain. As a result, the system stumbles ahead, becoming more and more disconnected from its initial goal.
It’s possible that this situation hasn’t yet reached its full absurdity, but take a look at the trajectory. According to estimates, one in four job applicants may be fraudulent in some significant way by 2028. This could include using AI to create fictitious experience, skills, or even conduct interviews through a proxy. In that world, the resume, which was already having trouble serving as a trustworthy indicator of talent, becomes virtually meaningless.
The document that used to be a candidate’s “here is who I am and here is what I’ve done” is reduced to an optimization problem, a search engine ranking challenge that is presented in a polished manner.
The technology itself is not the root of the issue. When used carefully, AI writing tools can actually assist a candidate in refining an argument or identifying a structural weakness. The issue is what happens when everyone uses the same tools in the same manner, which is precisely what has happened. The same action verbs, the same assertion of being results-driven and collaborative, and the same bullet-point architecture are all things that hiring managers report becoming increasingly uniform.
On a standardized document, culture-fit—which research indicates contributes significantly to turnover decisions—is almost undetectable. On paper, a new hire who says they are enthusiastic about cross-functional alignment might mean something quite different in real life.
It’s difficult to ignore what is lost during this filtering process. The applicant who worked for three years to turn around a failing department at a small business, doesn’t know the proper ATS keywords, and whose resume was never going to score a ninety-two on some system’s internal rubric is now statistically invisible to a large segment of the job market.
In the meantime, someone whose resume was designed to win a keyword match rather than a job is subtly appearing on the algorithm.
It is still genuinely unclear what will happen next. In an effort to measure what the resume was always meant to measure but is increasingly unable to, some companies are already shifting toward skills-based assessments, structured video submissions, and data-backed culture alignment tools. This has genuine momentum and makes sense. However, the vast majority of hiring infrastructure is still based on a document format that dates back to a time when typewriters were regarded as cutting-edge office equipment.
The resume won’t disappear overnight. It is surrounded by too many decades of habit and institutional inertia. However, in the areas that are most important, it is already functionally dead as an accurate, human portrayal of a person’s identity and abilities.
It has been replaced by a type of document that is designed for machines, filtered by machines, and increasingly composed by machines. The human being remains somewhere in the midst of all of this. Whether the procedure is still intended to locate them is the question.

