The Entry-Level Job Isn't Disappearing Because It Should

In July 2025 my flight out of California was delayed two hours. It was Friday, the Fourth of July, I was heading back to DC after visiting family, and the airport was pure chaos. By that summer, I was an airport regular, traveling to Chicago every Saturday for my weekend part-time MBA program for the prior year and a half. Despite being an extrovert, my flights had become my rare window of peace. I'd slap my massive Apple headphones on (courtesy of my old job at Spotify), pull the brim of my baseball cap over my eyes, and I did not engage with a soul.

But Fourth of July travel will throw off anyone's routine.

I ended up headphones off, deep in conversation with the man sitting next to me on the plane. He was flying back from a business trip and asked what I did. By habit I said I was unemployed. (Saying I was building a startup still didn't come naturally then.)

We got to talking about both having a background in engineering and at some point he mentioned his son had just graduated with a Masters from the University of Maryland. He was worried for him with this recent AI push. He said it felt like no one was hiring new grads.

Back when he was starting out, the degree was enough to get your foot in the door. And when I graduated, my experience was similar to his. I had my first job at Lockheed by the end of my first semester senior year. A full semester before I graduated. And I wasn't even close to being the strongest student. I could show you my transcripts but please don't ask.

We agreed that learning on the job is a hallmark of your first gig. The expectation used to be that you obviously didn't know what to do yet, but you were someone willing to figure it out. In my case, I knew basically nothing about submarines when I graduated college. A year later, was I an engineer for an advanced submarine warfare program? You bet.

Now companies want people who already have done the job. But how many new grads do you think have that experience? It's not like there's a 300-level submarine warfare class you can take.

It was approximately at this point that I remembered to mention "Oh wait, I'm not unemployed. My cofounder, Alison, and I are building a company around this exact issue." He asked if he could send our site to his son and give him my information. Obviously I said yes. I think technically when building a business I would be the one who is supposed to bring that up a lot sooner but luckily he took the initiative.

What stuck with me about that conversation was who it came from. Older. Engineer. Defense Industry. Currently employed. Not usually the person I'd expect to see this the same way we do. He was someone who played by the rules, built a career inside the system, and still could see clearly that something structural had changed. The doors he walked through for his generation, and that were still open for mine, were closing on his kid's.

And yet the dominant response from companies has been to make the problem worse. Entry-level job postings in the U.S. have dropped roughly 35% since early 2023. A 2026 survey of nearly a thousand business leaders found that by the end of next year, almost half expect to have eliminated entry-level hiring entirely. The stated reason is AI. Why pay someone to do work a model can handle faster and without the learning curve?

There's an obvious counterargument here: if you stop hiring junior people, where exactly do senior people come from in ten years? Is the plan that someone else trains them and you just poach the survivors? It's a real problem. Just not the most interesting flaw in the logic.

Lucky, Not Strategic

The assumption is that the work companies gave their entry-level employees was a reasonable representation of what those employees were capable of doing.

But it's not. It's always been a representation of what's low-risk enough to delegate.

Someone had to handle the repetitive, lower-complexity work. Junior employees were the safest, cost effective bet. It's a risk management decision. You don't hand a high-stakes problem to someone unproven. So you hand them the routine stuff and see what happens.

And what happened, over and over, is that people did more than what was asked. Because they were adults with their own ideas, instincts, and ways of seeing problems. They spotted patterns that the people embedded in the system had stopped noticing. They questioned processes that everyone else had accepted as given. Some of them independently identified problems and started building solutions that nobody had assigned them to build. Because they saw the need and figured they could help.

Hiring, at scale, was not well designed to develop talent, but it still produced results because people brought initiative. Companies have been accidentally benefiting from exactly the kind of thinking they're now deciding they don't need.

Societally, we've been lucky, not strategic.

And now the logic goes: AI can handle the tasks we were giving entry-level people, so we don't need entry-level people. That logic only holds if you believe that what someone was assigned is all they were ever going to contribute. That's evaluating based on the output of a system that systematically underinvested in these people.

It's the same pattern Alison wrote about: if you treat people like machines, of course you'll conclude a machine can replace them. You designed the comparison.

What's Left When the Grunt Work Goes Away

So look at what AI actually removes versus what it leaves.

AI handles the grunt work. The rote, repeatable, low-complexity tasks that junior people were assigned because it was the safest starting point. If that work goes away, you've removed the thing that was sitting between them and the work that actually matters.

What's left is the part that was always the real value: people who still question the system because they haven't absorbed the "that's how we've always done it" mentality. And today, they could be paired with tools that let them learn faster, push their thinking further, and execute on ideas instead of spending their first years just trying to prove they deserve to be in the room.

And you get the other side of that equation too. Senior people with actual bandwidth. Not buried under the operational work that used to consume them. But available to mentor, to invest in the people around them, to engage with a different way of thinking instead of just handing someone a task list and hoping they figure it out before they burn out.

I've been on both sides of this. I've been the junior person thrown into real work with no support, learning fast but burning out because the infrastructure to develop me wasn't there. And I've been the more experienced person who invested in someone junior and watched it make both of us sharper.

Mentorship, good mentorship, is a two-way exchange that forces you to reexamine your own assumptions.

Yes, AI can make individuals more productive. That's obvious. The less obvious part is what happens at the team level. A Harvard and Procter & Gamble field experiment found that teams with strong dynamics who also use AI produce three times more breakthrough solutions than those without both. Not marginally better. Three times. AI plus a team that genuinely thinks together, where different perspectives actually get heard and the conditions exist for people to challenge each other, is a categorically different kind of output.

We have never consistently had the conditions to test what entry-level people could do when given real support, real tools, and a team structure that actually values what they bring.

AI is creating those conditions for the first time. And the prevailing response is to decide we don't need entry-level people anymore. That's actually insane.

The Math Doesn't Check Out

Unfortunately, we don't get the satisfaction of there being a single villain to blame. The companies cutting entry-level roles are responding to what the system has shown them. That junior employees do routine work. And AI can do that same work faster. The math seems straightforward. But it is just measuring what people did when that was all they were asked to do and concluding that's all they were capable of doing.

It's not intentionally malicious or illogical. It's a mismeasurement that's been compounding for a long time, and this is the moment where it either gets corrected or gets locked in. Companies that recognize the difference between the task and the person will build teams that compound. Companies that don't will keep optimizing for the output of a broken system.

The entry-level job isn't disappearing because it should. It's that enough of the people making decisions have been inside the system so long they stopped questioning it. It's easy to do, honestly. Deciding that the way you've been doing something is working well enough and not worth reexamining.

Headphones Off.

For me, headphones-on was a decision I'd made a long time ago and never revisited. When flying is just a commute, you optimize for silence. But that flight was the first time I didn't feel fully delusional for the scope Alison and I were taking on. Someone with a completely different career, a completely different generation, who I never would have expected to care about the same things, saw exactly what we were seeing. I would have missed it entirely if my flight had been on time and I'd kept my headphones on like I always do. I'm starting to think I should stop wearing my headphones on flights altogether.

When I finally made it back to DC that night I texted Alison:

I am an extrovert. I genuinely love talking to people, hearing how other people see things. And I'd been shutting that off every flight because the normal expectation is you don't talk to the person next to you. But that's just not how I operate at my best. It's not for everyone, we all have our ways of engaging in different perspectives that work for how we operate. But regardless of the method, other people's perspectives matter. Entry-level, retired, whatever.