Computer science enrollment just took its biggest hit in years, and according to a Hacker News piece making the rounds, that panic might be the most contrarian buy signal in tech right now. Undergrad CS dropped more than 8 percent last year. Graduate enrollment fell 14 percent. Recent CS grads have a higher unemployment rate than philosophy majors, which is a sentence nobody expected to write in 2026.
What stands out here isn’t the doom. It’s the gap between the headline numbers and what’s actually happening underneath.
The numbers tell two stories
Yes, fresh CS grads are struggling to land jobs. But Hacker News flags a detail that gets buried in most coverage: CS majors have low underemployment. Translation, they’re not stocking shelves or pulling espresso shots. Nearly half of philosophy majors are. CS grads are holding out for actual engineering roles, and the ones who land them still significantly outearn their peers.
The unemployment rate is partly a function of being picky. That’s a very different problem than “the field is dead.”
Meanwhile, demand for mid and senior engineers is climbing across the industry. The ladder hasn’t collapsed. The bottom rung got chewed up.
Why the bottom rung disappeared
Anthropic co-founder Jack Clark said roughly 90 percent of the company’s new code is now AI-generated, and warned that “the value of more junior people is a bit more dubious.” That’s the quote that should be tattooed on every CS department’s strategy doc. Coding bots are now great at exactly the kind of tickets a junior used to grind through to learn the craft.
The pipeline that used to turn graduates into senior engineers got shorter and steeper. You don’t get five years of CRUD work to figure out how systems break. You’re expected to operate at a higher level from day one.
The split inside CS departments
Professors are visibly disagreeing about what to do, and the disagreement is the interesting part.
- Embrace the bots: Carnegie Mellon’s Michael Hilton keeps rewriting his curriculum and tells students to use AI to code. His line: “the things I taught three years ago are not the right things to teach today.”
- Go back to paper: Bard’s Valerie Barr now runs her intro class mostly on paper, the way she taught in the 1980s. Her argument: “You cannot make effective use of AI tools if you don’t know something about what you’re asking the tools to do.”
Both are right, and the split points at a bigger fracture. Is CS about producing software developers, or about teaching the computational theory underneath? As coding gets automated, those two paths are pulling apart fast. MIT’s AI major, launched in 2022, is already the second-most-popular on campus. Specialized AI tracks are eating the generic CS degree.
What this means right now
For anyone making bets in this space, a few things follow:
- For students: Don’t pick CS because it’s a safe paycheck. Pick it because you want to understand systems deeply. The generalist “learn to code” pitch is dead. The specialist pitch (ML research, security, distributed systems, AI infrastructure) is healthier than ever.
- For hiring managers: The mid-level shortage is your own fault. If you stop hiring juniors, you stop manufacturing seniors. In three years that math gets ugly.
- For founders: The bar for “technical co-founder” just moved. You don’t need someone who can ship a CRUD app. You need someone who can architect systems an AI will build and audit what it produces.
- For educators: Pick a side. Hybrid curricula that half-heartedly teach syntax while gesturing at AI tools will produce graduates who’re good at neither.
The Economist told students last week to “forget Python, study Plato.” That’s a cute headline and terrible advice. The AI revolution is making more of the economy run on software, not less. Someone has to understand how that software actually works when it breaks at 3 a.m.
The contrarian play isn’t fleeing CS. It’s going deeper into it while everyone else flinches. Full piece available at the original source.