The Hiring Market Is Officially Broken

Ken Schumacher ran engineering interviews for a tech company. Candidates were acing his coding tests — not because they were talented, but because the exercises got posted on Glassdoor and everyone practiced them cold.

Then applicants started using AI chatbots and teleprompter tools to fake their way through live interviews. According to The Atlantic, Schumacher now runs a startup using AI to detect AI cheating in job applications. An entire new industry has emerged just to verify that job applicants are real people.

Mitchell Hoffman, a labor economist at UC Santa Barbara, told The Atlantic that technology generally improves job matching efficiency. But AI is doing the opposite. Kathleen Creel, a philosopher and computer scientist at Northeastern University, called it "AI-on-AI crime" — employers deploying AI screening tools, candidates deploying AI application tools, and the actual human signal disappearing in the crossfire.

A Columbia Business School paper cited by The Atlantic described the outcome bluntly: AI tools are "compressing" and "homogenizing" resume content. Everyone looks better. Everyone looks the same. Hiring managers can no longer tell who actually knows what they're doing.

Silicon Valley has a name for it: signal collapse.

What Mainstream Coverage Is Getting Wrong

Most coverage of this story frames it as a fairness issue — AI democratizing access to opportunity, just with some rough edges. This framing misses the point.

This is not democratization. Democratization would mean more people getting jobs based on actual merit. What's actually happening is the opposite: meritocracy is being undermined because the tools that were supposed to level the playing field are instead rewarding people who are best at gaming AI systems. That's not the same as being good at the job.

The worker who actually has the skills but doesn't know how to prompt ChatGPT into a killer resume loses to the worker who's mediocre at the job but excellent at AI-assisted self-marketing. A system where actual abilities matter less than the ability to prompt an AI is a bad deal for talented people who don't game systems.

Meanwhile, in Healthcare: The Numbers Are Real

While the hiring market descends into chaos, the healthcare sector is reporting hard numbers from AI deployment.

Hospital for Special Surgery (HSS) in New York, an academic medical center focused on musculoskeletal health, has deployed AI agents to handle insurance claims processing. According to MIT Technology Review, those agents now complete 1,100 claims per month. The appeals stage dropped from 45 minutes to five minutes. The success rate on appeals went from 65% to 100% over nine months. Dr. Ashis Barad, MD, chief digital and technology officer at HSS, reported these operational results.

The World Health Organization has warned that global healthcare worker shortfalls will hit 11 million by 2030. A KPMG survey cited by MIT Technology Review found that 68% of healthcare providers have already deployed AI agents into their workforce.

The difference between healthcare AI and hiring AI is accountability. A hospital can measure claim processing time. It can measure appeal success rates. It can track patient outcomes. The results are verifiable and the incentives are aligned — faster claims processing saves the hospital money and reduces staff burnout.

In hiring, the incentives are misaligned. Candidates want to get hired. Employers want to screen efficiently. AI tools serve both of those goals in ways that produce no improvement in actual job matching quality.

The Sponsored Article Problem

One flag worth raising: the MIT Technology Review piece on healthcare AI is sponsored content — published in partnership with Ema, a company that sells agentic AI products for healthcare. The HSS results it cites may be entirely accurate, but that context matters. Sponsored journalism is not the same as independent reporting. The numbers look good. They should be verified independently before anyone treats them as settled fact.

MIT Technology Review is a credible outlet. But credible outlets running sponsored content should disclose it prominently — and readers should apply appropriate scrutiny.

The Same Technology, Two Very Different Outcomes

AI produces gains when it's applied to structured, measurable, repetitive tasks — processing insurance claims, triaging paperwork, flagging anomalies in data. It destroys value when it's applied to signal-dependent, judgment-based tasks — evaluating human potential, assessing fit, reading between the lines of a resume.

Hospitals are deploying AI to handle the administrative grind so doctors can do doctoring. That makes sense.

Companies are deploying AI to screen the human judgment out of hiring so algorithms can do the deciding. The evidence suggests this doesn't work.

What This Means for Regular People

If you're job hunting right now, you are competing in a broken market. The incentive to use AI tools to polish your application is real — because everyone else is doing it. But the result is a system where your actual abilities matter less than your ability to prompt an AI.

If you're a patient, the AI-in-healthcare story is genuinely promising — IF the results hold up outside of sponsored case studies and IF regulators keep pace with deployment.

The outcome depends entirely on what you're asking the technology to do, who benefits from the outcome, and whether anyone is honest enough to measure the results.

Right now, in healthcare, someone is measuring. In hiring, nobody is — and the damage is already done.