Everybody Passed the Test

I once spent an afternoon trying to make a form forgive me for having a middle initial.

This was before the machines learned to write little love letters to human resources. The form did not want Henry Charles Chinaski. It wanted Henry Chinaski, exactly as some clerk had decided Henry Chinaski ought to appear. I removed the initial. It accepted me. For a moment I felt the kind of victory usually reserved for prisoners who have learned how to make a spoon out of a toothbrush.

Then I remembered I still needed the job.

That is the part the new hiring machinery has polished into invisibility. A person is looking for work. He is not looking for a game. He is not hoping to spend Tuesday afternoon discovering whether a computer prefers a percent sign to the word percent. He wants somebody to look at the years he gave away, the bad bosses he survived, the things he learned to do when nobody was around to show him, and decide whether he might be useful.

Instead he is kneeling before a slot machine with a résumé.

He feeds in verbs. He adds keywords. He cuts the second page because some invisible judge has an aversion to paper. He asks one machine to rewrite the document so another machine will find it appealing. The first machine makes him sound like a confident young executive who has never had a sick relative, a lousy apartment, a gap in employment, or a reason to use the phrase “I handled whatever came through the door.” The second machine measures the result and gives him a score.

Seventy-two.

Eighty-four after he changes “percent” to “%.”

Ninety-one if he calls himself a strategic cross-functional something-or-other and pretends he did not spend six years keeping a real operation from catching fire.

Everybody in this arrangement is trying to get through the same locked door. The applicant wants an interview. The recruiter wants five minutes without two hundred nearly identical applications arriving like pigeons with business degrees. The manager wants somebody competent before the quarter dies screaming. So each of them buys a machine.

Then the machines begin corresponding with one another while the people wait outside.

This is being called an AI doom loop, which is a fine name if you enjoy giving a fever an interesting brand identity. I call it an office full of mirrors. The applicant asks a bot what a hiring bot wants. The bot replies with the smooth gray porridge it has learned from every other applicant asking the same thing. Hundreds of people receive slightly different versions of the same advice. Hundreds of résumés arrive bearing the same bright adjectives, the same obedient cadence, the same little claims of passion and impact and leadership.

The recruiter looks at the pile and sees a mass-produced lie.

Maybe it is a lie. Maybe the applicant used the machine because rent is due and the job description was written by a committee of escaped filing cabinets. Maybe she is a brilliant designer who has been reduced to arguing with a score because somebody told her the score decides whether a human being will ever see her work. Maybe he is fifty-two and has spent his adult life solving actual problems, but the system wants the right sacred words from 2026 and has no field for did not panic when everything went wrong.

So the recruiter buys more filtering.

There it is. The loop does not begin with laziness. It begins with exhaustion. Every person in it has been handed too much noise and a smaller piece of the day. The machine promises relief. It takes the noise, processes it, and returns it with a tie on.

I do not blame the applicant for using the thing. Job hunting has always had the dignity of a bad blind date arranged by a landlord. You polish yourself. You omit the parts that make you human. You sit across from a stranger and try to make thirty years of being alive sound like a useful response to “Where do you see yourself in five years?” A machine that helps you get the commas right is not the enemy.

The enemy is the little superstition that more filtering makes judgment better.

One company tested automated rankings against people it had already hired, people who had been doing fine in the job for months. In two tests, the actual hires did not make the machine’s shortlist. The useful people were down in the basement with the bad formatting, apparently, while the system congratulated itself upstairs.

I have seen that kind of intelligence before. Every bad supervisor has it. He notices the worker who fills out the paperwork cleanly and misses the one who knows which truck will fail, which customer is lying, which new kid needs ten minutes of instruction before he makes a week of trouble. The first kind of knowledge fits in a box. The second kind costs time, attention, and the uncomfortable risk of looking somebody in the eye.

Companies say there are too many applications to read. I believe them. They have built a labor market where applying for a job is free, desperate, instantaneous, and often pointless. Of course the pile grows. A man who receives no answer after eighty applications does not become serene. He sends eighty more. Then a hundred. Then the machine writes them faster. The company sees the flood and installs a better drain.

Nobody is hiring. Everybody is sorting.

The strange thing is that people still get hired the old way. Someone remembers them. Someone sees their work. Someone writes a cover letter that sounds like it came from a person with a pulse, or meets a former colleague for coffee, or is recommended by the woman who knows they can be trusted with the keys. The system calls this networking, because it needs a bloodless word for the fact that trust is still built between people.

That does not make it fair. Plenty of good workers do not have the right circle, the right city, the right old boss willing to take a call. But it does tell you what the machines cannot solve: a job is not a bag of keywords, and a person is not a résumé with the serial numbers filed off.

Somewhere tonight, a job seeker is changing one word to another word and watching a score rise. The room is quiet. The bills are not.

On the other side of the screen, a recruiter is opening another stack of perfect applications and wondering why none of them feels alive.

They are both being tested.

Neither of them wrote the test.

Source: AI Use in the Job Market Is Creating an Infinite Doom Loop