The Irish side of my family has a slightly embarrassing habit of finding other people's misfortunes hilarious. A kid belching in the middle of a funeral would get big laughs in the retelling. It's not ridicule, it's more like a sympathetic acknowledgment of life's absurdity.

I mention this because it's the best way to frame my recent experience being interviewed by AI. Absurd and funny.

It went like this: I clicked a link, and a chatbot initiated our session. We got through some qualifying questions pretty easily. Then it asked how I typically make business decisions — about media mix, budget, metrics — for digital campaigns. Campaign management has never been part of my experience, let alone my career, which I explained. It asked the same question again. I said I'd already answered this, and by the way, those responsibilities don't appear anywhere in the job posted. It asked a third time. At that point I was feeling ignored, so I said I didn't think we could continue our date.It asked again. I closed the window.

My experience turned out to be a pretty good microcosm of hiring in 2026. Neither side is fully engaged anymore, both employer and candidate have outsourced the process to AI, and what's left looks like speed dating: fast, transactional, and kind of superficial.

1. AI versus AI

In his excellent recent article in The Atlantic, Ian Bogost takes an honest look at AI dynamics in recruiting. He points out that when candidates use AI to respond to job openings, they are not "cheating," they are showing a perfectly reasonable response to a dynamic they didn't create. If employers deploy AI with candidates, candidates deploying it back isn't fraud, it's leveling the playing field.

The numbers back him up. According to HackerRank, 14% of candidates openly admit to using generative AI on tech assessments, and 83% say they would use it if they thought employers couldn't spot it. And here's the irony: One report found 61% of candidates flagged for using AI would have advanced without detection. In other words, they weren't just getting away with it, they were moving forward.

Bogost's phrase for this is fake it 'til you fake it, or use the tools at your disposal — including the tools used by the employer — to reach the next level, where you will likely need the same tools again. Nobody in this chain is lying, technically. They're just following an incentive structure that was created for them.

2. Swiping left

Meet Wafa Shafiq, who applied to a marketing role and was contacted within minutes by Alex, who scheduled her interview for the next day. Turned out Alex wasn't a person, Alex was an AI recruiting agent, the kind now used by a number of big companies. Shafiq posted about the experience, calling the interaction "really dystopian."

I once received a message from an AI screener that told me to call it back. I tried three times and got dead air each time — not a bad connection, not hold, just nothing. A day later, I received an email from a human recruiter asking if I had completed the call. When I explained the connection failure, I was told to try again. Same result.

When rejecting a candidate costs nothing, and bad screening decisions can go unnoticed, there's no limit to how many people can be ruled out. It's dating app behavior applied to recruiting: everyone thinks their "unicorn" is one more swipe away. But more options don't make hiring managers more discerning. They just make rejection easier.

Years ago I worked briefly as a gallery assistant in San Francisco. The gallery director, Ed, was an older salesman with a New York accent, a real-life Willy Loman. The gallery sold old master prints and drawings. One afternoon a customer became very interested in a 17th-century Blaeu map. Ed trotted me out as the unofficial research assistant to explain the print, which I did, then mentioned we had a few similar ones available. The moment the customer looked away, Ed turned to me and stage-whispered a single word: "No."

I was mortified. He later explained that more options don't close a sale, they create doubt. Give a customer too much choice, and they have to "think it over."Alvin Toffler, author of Future Shock, coined a great word for this phenomenon: overchoice. Past a certain point, more options become a burden, not a benefit. A hiring manager looking at hundreds of AI-screened, pre-sorted candidates may actually be less discerning than one looking at six. When abundance becomes the default, the definition of the "right" candidate starts expanding with it.3. What's new is old

Abundance shows up in other places, too.Have you noticed the weird anachronisms creeping into job descriptions? A new type of requirement is starting to show up: three to five years of experience with a skill that's less than two years old. Take AEO, or Answer Engine Optimization: creating content AI can present as a direct answer, rather than a ranked result in search (which still happens, but seems almost quaint).

I've come across at least six companies that want AEO as a required skill. The catch is that two years ago, it didn't really exist, at least not in its current form. Nobody has three years of experience doing something that's two years old, including the people writing the job description.

When adding another requirement costs nothing, there's little reason not to add it. The job description turns into a fantasy shopping list, with every new skill required, even when the skill itself is still being invented. Companies are creating their unicorns in real time, and candidates are using AI to turn themselves into those unicorns.

4. The feds

If you want to know what discernment in hiring is supposed to look like, there's a federal document for that (for now). It's known as UGESP, the Uniform Guidelines on Employee Selection Procedures, and it lives at 29 CFR Part 1607.

Among other things, UGESP says that if a selection procedure has an adverse impact on a protected group, the employer needs to show that it’s valid. The "four-fifths rule" flags a potential problem when one group's hiring rate falls below 80% of the group with the highest rate. If a test screens people out, the employer should be able to show that it actually predicts job performance.

So, in theory, there's a documented, federally compliant record showing how asking me to explain my media-mix methodology four times in a row predicts my job performance. Right? Riiiiight? 

5. Into the void

Based on what I've read, many large employers don't build their own hiring algorithms, they buy them from vendors. A recent study examined 3.4 million real applicants across 156 employers, all screened by algorithms from a single vendor. The researchers identified something they called "algorithmic monoculture": when enough employers rely on the same system, independent decisions can start to become correlated. That doesn't mean every rejection follows you from job to job, but it does suggest some decisions may be less independent than they look.

One finding: 10% of candidates who applied to four different companies, all screened by the same vendor, were rejected by all four. If each company's decision were truly independent, fewer people would be rejected across the board. The researchers also estimated that in a monoculture situation, a candidate may need about 25 applications to have a 99.9% chance of being recommended at least once by the algorithm.I've read many accounts from job seekers who say it feels like their resumes are just vanishing into a void. That's not paranoia. Some of those resumes are being sorted by the same handful of algorithms over and over, with each rejection presented as a new decision.

6. That wasn't fast

Most AI tools are marketed with the same promise: speed. And for a historically slow process like recruiting and hiring, what could be better?

But recent data points in the other direction. According to a 2026 survey from GoodTime, 90% of companies missed their hiring goals in 2025, 60% saw time-to-hire get worse rather than better, and only about one in nine managed to speed anything up.Meanwhile, the Bureau of Labor Statistics reported 6.9 million open roles in February 2026 and only 4.8 million actual hires, the lowest hiring rate since April 2020.

Something may have gotten faster, but it isn't hiring.

7. Do as we say

But let's get back to job seekers using AI.

I recently noticed Adobe includes this on its careers page:

"Our interviews are designed to reflect your own skills and thinking. The use of AI or recording tools during live interviews is not permitted unless explicitly invited by the interviewer or approved in advance."

Then later:

"At Adobe, we empower employees to innovate with AI — and we look for candidates eager to do the same."

That's an interesting contradiction. What if I want to innovate with AI to get an interview or a job, does that count? After all, the tool banned during the interview is at the heart of the role.

I don't mean to single out Adobe. They are by no means unique in this. But they and companies like them are asking you to prove you don't need the tool they expect you to use.

If hiring teams only knew what AI is saying about them and their job postings. My Irish family would find the whole thing hilarious.