Let me tell you what looking for a job feels like in 2026, because the numbers this week stopped me cold.
The average job opening in America now attracts 243 applications. Three years ago — the first quarter of 2022 — it was 89. That’s according to Greenhouse, the hiring platform, and reported in Morning Brew’s weekend edition. Nearly triple the competition, in three years. If you’re job hunting right now and it feels like shouting into a void, you’re not imagining it. The void has 242 other people in it.
And the void is increasingly staffed by robots. Ninety percent of U.S. employers now use AI screening tools, per a Stanford University study. Nearly two-thirds of American job seekers say they’ve been interviewed by an AI bot, per Greenhouse. The first conversation you have with your next employer may not be with a person at all — it may be with software, deciding in milliseconds whether a human ever learns your name.
Here’s the part that should make everyone uncomfortable: the machines are not as smart as the managers think. Sixty percent of hiring managers say they can identify AI-generated job applications — but only about 25 percent actually can, according to a Software Finder survey. So we have an arms race of artificial intelligence on both sides: candidates using AI to polish résumés and cover letters, employers using AI to filter them, and hiring managers confidently misjudging both. It’s a hall of mirrors, and somewhere inside it, a real person with real skills is trying to get a real job.
Companies are responding the way institutions always do when the system stops working: they’re going around it. Morning Brew’s Matty Merritt reported this week that businesses are increasingly relying on “work trials” before hiring — asking candidates to complete an unpaid mock task off-site that mirrors the job’s duties, or to spend a week in the office with the team. About two-thirds of companies now use some form of skill-based hiring for entry-level roles, per a 2025 survey by the National Association of Colleges and Employers.
There’s genuine wisdom in the work trial. A résumé is a story someone tells about themselves; a work trial is a demonstration. For a young person without connections or credentials, showing what you can do beats listing where you went to school. Skills-based hiring is, at its best, more democratic than pedigree-based hiring.
But let’s be honest about the shadows, because Merritt’s reporting didn’t flinch from them. Work trials can be a barrier to employed workers who can’t take a week off to audition for a job they don’t have. And they can turn exploitative when they’re unpaid — though the law generally requires that candidates be paid if their trial work benefits the company. There’s a fine line between “show us what you can do” and “do free labor and we’ll think about it.” Every job seeker should know where that line is, and every employer should be ashamed to cross it.
Now, the counterintuitive good news — and it’s genuinely good. A Gusto report released Thursday morning, covered by Axios, found that very small businesses that adopt AI actually grow their employee ranks faster than those that don’t. Read that again: the smallest employers, the ones using AI the most eagerly, are hiring more humans, not fewer. The story that AI simply destroys jobs turns out to be too simple. In the hands of a five-person company, AI handles the drudgery — the scheduling, the bookkeeping, the first-draft everything — and the humans get hired to do the human work: selling, caring, creating, deciding.
The Bureau of Labor Statistics’ latest projections add the long view. Health care and social assistance will see the most job growth over the next decade, driven by an aging population, while overall job growth is projected to slow. Zoom in and the single biggest number is almost poetic: 847,300 new home health and personal care aide jobs projected nationally between 2025 and 2035. Nearly a million jobs caring for our elders and neighbors — jobs that no robot interviewer can do, jobs that no AI can fully replace, because they require the one thing machines don’t have: a human presence in the room.
And the near-term news is solid, too. Employers added 162,000 jobs in August — far better than economists expected, per the Wall Street Journal. Hiring, in Finimize’s phrase, “roared back,” with the most jobs created since March.
It’s not enough to just optimize our résumés for the robots and hope for the best. It’s not enough to treat the hiring process as a game to be gamed — AI cover letters versus AI screeners, each side escalating, nobody actually meeting. We must listen to what the 243 applications are telling us — that standing out now requires proof, not promises — learn the new rules without surrendering our dignity to them, and contribute to a hiring culture that remembers what’s being hired: not a keyword match, but a person.
For job seekers, the practical wisdom writes itself. Build a portfolio, not just a résumé. Do the work trial when it’s fair; walk away when it’s exploitative. Learn to work alongside AI rather than competing against it — the Gusto data suggests the workers who thrive will be the ones who treat AI as a power tool, not a rival. And consider where the growth actually is: in care, in health, in the human work that an aging country desperately needs.
For employers, the message is sterner. Your AI screener is turning away talented people — the Stanford research says so. Your hiring managers can’t spot AI-generated applications nearly as well as they think. The work trial is a better filter than the keyword filter, but only if you pay people for their time and respect the ones who can’t afford to give you a free week. The companies that figure out how to actually see humans in a sea of 243 applications will win the talent war. The ones that let the robots decide will get exactly the workforce they deserve.
Two hundred forty-three applications. One robot interviewer. Zero excuses for forgetting that behind every application is a person trying to build a life. The technology changed. The humanity didn’t — or at least, it shouldn’t have. That’s up to us.
Written from the September 11–14, 2026 editions of Morning Brew, Axios Markets, Finimize, and WSJ What’s News. Statistics are as reported by those outlets and the organizations they cite (Greenhouse, Stanford, Software Finder, NACE, Gusto, BLS); my reflections are my own take.
