For the past decade the tech industry had one message: you don't need a CS degree. Learn to code in a bootcamp. Watch YouTube tutorials. Build projects. Get hired.
And honestly? They were right. You didn't need to understand how a compiler works to build a React app. You didn't need discrete math to ship a CRUD API. The industry needed hands that could type, not minds that could reason about computation.
That era is over.
What changed
AI writes the code now. Not perfectly, not always, but well enough that "I can write code" is no longer a differentiator. The hands that type lost their edge.
What AI can't do is decide what to build, why an architecture will fail at scale, or when a solution is subtly wrong. That requires understanding the layers underneath. The stuff a CS degree actually teaches. Operating systems, algorithms, distributed systems, computational complexity - the things bootcamps skipped because they "weren't practical."
Turns out they're the most practical things left.
The inversion
Here's the irony. When coding was manual, you could get away without theory. You'd hit a wall sometimes, Google the answer, move on. The feedback loop was slow enough that pattern matching worked.
With AI the feedback loop is instant. You can generate ten architectures in a minute. But if you can't evaluate them - if you don't understand why one will deadlock under load and another won't - you're just picking randomly from a menu you can't read.
The people who skipped fundamentals could survive when the bottleneck was typing speed. Now that the bottleneck is judgment, they're exposed.
What a CS degree actually gives you
Not the syntax. Not the frameworks. Those expire every few years anyway.
It gives you mental models. You learn how to think about problems before thinking about solutions. You learn that every design decision is a tradeoff. You develop intuition for complexity - not just Big O on a whiteboard, but the real kind. The kind that tells you this "simple" feature will quietly destroy your database in production.
You don't get that from a tutorial. You get it from grinding through proofs you hated at the time and only appreciated three years later on the job.
The new split
The industry is splitting into two groups. People who use AI to generate output, and people who use AI to build systems. The first group will be large, cheap, and replaceable. The second group needs to understand what's happening under the hood.
A CS degree isn't the only path to that understanding. But it's the most structured one. And in a world where AI handles the surface-level work, depth is the only thing that still compounds.
The degree isn't dead. It just finally matters for the right reasons.