Why Employers Made Hiring So Frustrating

Looking for a job has become exhausting.

Applications disappear into automated systems. Interviews stretch across weeks. Technical assessments multiply. Rejections arrive without explanation, when they arrive at all. Candidates tailor their résumés to job descriptions they suspect were written by software, then submit them to screening systems they do not understand and may never have agreed to be evaluated by. By the time a human being enters the process, if one ever does, the applicant may already feel less like a prospective employee than a record being processed.

That frustration is legitimate. The use of artificial intelligence in hiring is now facing serious legal and regulatory scrutiny. Federal agencies have warned that algorithmic tools can unlawfully screen out qualified applicants with disabilities, and a closely watched federal lawsuit alleges that automated applicant-screening software discriminated against job seekers based on characteristics including age, race, and disability. The company involved disputes those allegations, and the courts have not resolved the case, but judges have allowed significant claims to proceed. (EEOC)

I personally know someone who believes they may have been affected by the kind of automated screening now being challenged. I cannot know whether that belief is correct, and neither can they. That uncertainty is part of the problem. When a rejection comes from an opaque system, candidates are left to wonder whether they were unqualified, unlucky, improperly filtered, or never meaningfully considered at all.

Employers should not dismiss that concern. A hiring process that cannot explain itself will eventually lose the confidence of the people subjected to it.

But there is another question that receives far less attention: Why did employers build a process that frustrates so many candidates in the first place?

From the outside, the answer can seem obvious. Companies became indifferent. Human resources departments added bureaucracy. Hiring managers forgot what it feels like to look for work. Technology made it easier to reject people at scale, so employers used it without much concern for what the experience felt like on the other side.

Some of that criticism is deserved. Companies do ghost applicants. Some leave positions posted long after hiring plans have changed. Some subject candidates to six rounds of interviews because nobody inside the organization is willing to make a decision. Others use automation without understanding how it works or whether its conclusions can be defended.

Yet that is not the whole explanation.

Hiring became more frustrating because employers gradually stopped trusting the signals on which hiring had long depended.

For decades, the process rested on a collection of assumptions that were rarely stated because they did not need to be. The person sitting across the table was the person named on the résumé. The answers offered during the interview belonged to that person. The experience described on the application reflected work they had actually performed. If an offer was accepted, the person who appeared on the first day would be the same person the company had evaluated.

Those assumptions were never perfect. People have exaggerated qualifications for as long as résumés have existed, and bad hires are hardly a modern invention. But the signals were generally reliable enough that employers could focus primarily on finding the strongest candidate rather than authenticating every part of the interaction.

Remote work changed that balance.

That change brought enormous benefits. Companies gained access to talent beyond commuting distance. Employees gained opportunities that geography had previously denied them. Teams became more flexible, and many organizations discovered that work they had assumed required physical proximity did not require it at all.

Remote hiring also removed many of the informal signals that had quietly supported trust. The shared room disappeared. Geography became more difficult to verify. An interview was reduced to a face, a voice, and a rectangular view selected by the candidate. None of that made remote candidates less capable or less honest. It simply narrowed what an employer could observe.

A more global labor market added another layer of complexity. Again, the problem was not that talent existed outside the United States. Exceptional engineers work everywhere, and many companies are stronger because they hire them. The problem was that location, employment status, tax obligations, security requirements, and the identity of the person performing the work all became more consequential at the same time they became more difficult to verify.

I saw that breakdown personally.

A company believed it had hired a United States-based employee to perform a technical role. The individual had interviewed for the position, represented their location and working arrangement, and was treated as the person responsible for the work. Over time, inconsistencies began to appear. Communication patterns were unusual. The quality and style of the output varied in ways that were difficult to reconcile with a single person. Availability did not always correspond with the hours the employee claimed to be working.

Eventually, the company discovered that the person it had hired was quietly sending the work to people in Pakistan.

The issue was not Pakistan. It was not offshore work, and it was not an argument against distributed teams. Companies knowingly work with overseas employees and vendors every day. Those arrangements can be productive, ethical, and entirely appropriate.

The issue was that the company had agreed to one employment relationship and was unknowingly participating in another. It believed it had evaluated, hired, and granted access to a particular individual. In reality, it no longer knew who was producing the work, who had access to company information, how many people were involved, or where the work was being performed. The employee had not merely exaggerated a credential. They had substituted an undisclosed delivery model for the person the company believed it had hired.

That kind of experience changes an organization’s assumptions. It does not make every future candidate dishonest, but it makes innocence harder to presume. Security reviews become stricter. Identity checks become more formal. Location requirements receive closer scrutiny. Managers begin asking questions that would once have felt unnecessary or even insulting.

A peer experienced a different version of the same collapse. His organization interviewed a candidate, completed the background check, extended an offer, and prepared for the new employee’s arrival. The company purchased and shipped a MacBook Pro worth roughly $3,000 as part of the onboarding process. The equipment was delivered, but the employee never began work and ceased participating in the process.

The company had followed its established procedures. It had interviewed the candidate, verified the information it knew how to verify, invested in the hire, and prepared the team. None of those steps prevented the outcome.

More recently, generative AI has weakened another set of signals.

I have participated in interviews where candidates logged AI assistants into the conversation. I have watched candidates pause after questions, glance repeatedly toward another part of the screen, and then deliver answers with a fluency that disappeared as soon as the discussion moved beyond the prepared response. I have been in interviews where someone appeared to be receiving direct coaching from outside the camera’s view.

Again, these candidates are not representative of everyone looking for work. Most applicants are honest. Most want a fair opportunity to demonstrate what they know. But organizations do not redesign hiring processes around the average candidate. They redesign them around failures they believe they cannot afford to repeat.

That is how controls accumulate.

A résumé becomes less trustworthy, so the employer adds a technical assessment. Candidates learn to optimize for the assessment, so the employer adds a live interview. AI makes the live interview less conclusive, so the employer adds screen sharing, identity verification, or another interviewer. Concerns about coaching produce more structured questioning. Concerns about identity produce additional background checks. Concerns about remote access produce stricter equipment, location, and security policies.

No single step appears unreasonable when viewed in isolation. Each one responds to a real failure or a plausible risk. Together, however, they create the process candidates now experience as needlessly burdensome.

The process did not become frustrating because one executive designed it that way. It became frustrating because precautions accumulated faster than anyone removed them.

Candidates then respond rationally to the system in front of them. If an automated screening tool may reject a résumé before a recruiter sees it, tailoring the résumé to the job description is sensible. If hundreds of applications produce only a few responses, applying at scale is sensible. If employers use AI to screen candidates, using AI to prepare an application can feel less like cheating than restoring balance.

The trouble begins when optimization makes the original signal less useful.

A résumé once represented a candidate’s best summary of their experience. Now it may be a machine-generated reflection of the job posting, carefully arranged to mirror the language an applicant believes another machine is searching for. The document may still be accurate, but its apparent fit no longer tells the employer as much as it once did.

The same problem appears in interviews. Preparation has always been part of interviewing, and there is nothing improper about using AI to study, rehearse, or organize one’s thinking beforehand. The distinction changes when the tool participates invisibly in the evaluation itself. At that point the employer is no longer measuring only the candidate’s judgment or knowledge. It is measuring an undisclosed combination of the candidate, the model, the prompt, and whatever outside assistance may be present.

Employers respond by adding another layer. Candidates interpret that layer as further evidence that companies do not trust them. They adapt again. The employer then concludes that the previous control no longer works.

The pattern becomes self-reinforcing.

Economists often describe markets as systems of signals. A degree signals education. A credit score signals repayment risk. A certification signals that someone has met an established standard. Hiring relies on its own collection of signals: résumés, references, interviews, assessments, portfolios, credentials, and professional reputation.

When signals are broadly trusted, transactions can remain relatively simple. When those signals become unreliable, the system compensates with verification.

That verification carries a cost.

It consumes time from candidates who have done nothing wrong. It requires managers to spend more hours interviewing and fewer hours leading. It delays hiring decisions, drives strong applicants away, and favors organizations large enough to support elaborate screening processes. It can also disadvantage candidates who are highly capable but less practiced at navigating the defensive machinery surrounding modern employment.

The burden falls most heavily on honest participants because dishonest participants are already willing to bypass the rules.

An applicant who uses AI secretly during an interview may cause the next hundred applicants to face a more intrusive process. An employee who misrepresents their location may make the next legitimate remote hire endure additional scrutiny. A person who accepts equipment and vanishes may lead an organization to delay shipping equipment or require more proof before onboarding.

The reverse is equally true. Every employer that ghosts a candidate teaches that candidate not to expect basic courtesy. Every unexplained automated rejection strengthens the belief that applications are disappearing into systems no one understands. Every questionable posting encourages applicants to spread their effort across more opportunities rather than investing meaningfully in any one of them.

Both sides learn from the worst behavior of the other.

This is where the argument can easily become a defense of employers, and it should not. A bad candidate experience does not become acceptable merely because fraud exists. Six interviews are not automatically justified because someone once cheated. Companies still have an obligation to design processes proportionate to the risk, explain what they are evaluating, protect applicants from discriminatory tools, and treat people with basic respect.

The legal scrutiny surrounding algorithmic hiring matters for that reason. Federal agencies have already warned that employers remain responsible when automated tools unlawfully disadvantage applicants, even when those tools come from outside vendors. The use of software does not transfer accountability away from the company making the employment decision. (EEOC)

Employers therefore face two obligations that are increasingly difficult to reconcile. They must verify that candidates are genuine without treating every candidate as presumptively dishonest. They must use technology to manage application volume without allowing that technology to make opaque or discriminatory decisions. They must protect their organizations without turning hiring into an investigation.

Many have not found the balance.

The result is a process that feels hostile from both directions. Candidates enter expecting automation, indifference, and rejection. Employers enter expecting exaggeration, optimization, coaching, and occasionally outright fraud. Each side begins with less trust than the interaction requires, then interprets the behavior of the other through that suspicion.

That is why hiring has become so frustrating.

Employers did not set out to create a process candidates dread. They responded to remote work, expanded labor markets, increased application volume, weakened signals, security concerns, and new forms of deception. Candidates did not set out to make themselves impossible to evaluate. They responded to opaque screening, low response rates, impersonal rejection, and a system that often seems designed to exclude them before anyone listens.

Each response makes sense on its own.

The accumulation does not.

The larger lesson extends beyond hiring. Every institution depends on some degree of trust. When trust is present, people can rely on judgment, discretion, and relatively simple processes. When it disappears, institutions replace it with forms, controls, audits, documentation, monitoring, and approvals.

Some of that process is necessary. Much of it begins as a reasonable response to a genuine failure. But process is a poor substitute for trust. It is slower, more expensive, and less capable of distinguishing between the honest person who deserves confidence and the dishonest person who caused the control to exist.

Hiring is now caught in that substitution. Employers increasingly optimize not only for finding excellent people, but for avoiding deception. Candidates increasingly optimize not only for presenting themselves honestly, but for surviving systems they believe may never evaluate them fairly.

Neither side would have designed this process from scratch.

They built it one rational decision at a time.