Computer Science, are you sure?
Do not study Computer Science anymore. Nobody needs you, what are you going to do after your studies? It’s the stupidest and hardest thing to get into today.
That’s stuff we, we as in CS students, and that includes me, put up with. For me it’s at least bi-monthly that I get into such a discussion. And as discussions are short or cut off, here’s my actual response.
And I have thoughts.
Computer Science is not about coding.
It never was. The same way astronomy is not about telescopes, cooking is not about recipes and music is not about instruments. A musician uses instruments to create art. The instrument is the tool that gets you to a result. A lot of people know how to play the piano, but that doesn’t make all of those people great artists. Because it is about so much more than plucking the strings. And if everyone in the world was given a guitar, we still wouldn’t have many more popular musicians, and the majority of people wouldn’t even touch theirs.
This is what’s happening to coding. Creating new technology and software has never been about writing code. Code is cheap, and it just got cheaper. But the cost of software hasn’t moved. Everyone in the world is being given the guitar, everyone could open Claude or Codex and start building.
software
code
three years
Either way, you are not some sort of mastermind wizard because you are telling me that no one is going to code in trying to tell me that „our jobs are gone“. In fact, I’ve known since 2023, and I’m serious when I say that coding never was a career goal for me.
Again, Computer Science is not about coding. And if you tell people that we really don’t do much coding in our ETH CS program, they are confused, asking what else we do then.
Back to foundational science
If Computer Science isn’t about coding, what else do we do then? It’s about foundational theory, transforming information, reasoning. It’s about breaking down complex systems, understanding how they interact, and finding elegant solutions. And the term Computer Science will move back to its roots, towards deeper thinking and mathematical insights, engineering at its core.
Other foundational science majors (maths, physics, etc.) are not getting all the grief and attitude that CS majors get, even though their skillset is just as affected as ours. We develop solutions. We use code to create those solutions, but it is about the solutions, not the code. The average vibe coder has produced an incredibly large number of lines of code and has solved an incredibly small number of true engineering challenges or problems we have in this world.
So Computer Science goes down, deeper into engineering, but on the other side, in regards to actual coding the “new” Software Development (again, Software Development ≠ Computer Science) goes up abstractions above the code. Let me explain.
Development hasn’t gotten easier
I don’t like the term vibecoding. It feels cheap, when in reality the software that you create through agentic engineering has the highest quality there could be, if done right. We will solve the worlds hardest challenges through agentic engineering, but that doesn’t mean engineering is dead.
The “if done right” part is the point. It’s easy to send a prompt and be stunned by the output, but now what? The thing people don’t realize is that 20% of work will get you 80% of the way, and it’s the final 20% that makes the difference, and that’s where 80% of the work lies. Developing software isn’t a one-off thing. People realize soon enough that paying $20 a month for a service you can trust is well worth it compared to developing and maintaining your own thing.
On an unrelated note, here’s to vibe coding
Not really connected to this topic of computer science, “everyone can vibe code” isn’t entirely true either. You can be good or bad at agentic engineering. Sending a prompt and scrolling on Instagam waiting for it to finish, is easy. But that way you are the bottleneck, which you don’t want to be. Spin up a 2nd agent. 3rd, 4th, up to the point where your mind can’t keep all of them in memory. But at that point your 1st agent is done, and you prompt it again. Co-working with models and orchestrating them to the point where your complex project is actually ready, that is hard.
one agent
mostly idle
four agents
mostly busy
And even amongst CS developers, using AI to help with the coding part of a project is getting lumped in with the word vibe-coding, and vibe-coding gets associated with easy, lightweight work. Work that doesn’t require much concentration, effort, or skill. And that so isn’t the case. It is hard creating a product even if you decide not to read a single line of code. I am convinced by this and the fact that anyone who says otherwise has never actually done it. Besides this, I still to this day read through every line of code in my PR., because that PR has my name on it and I want to know what’s going on (and partly because I simply enjoy code).
All of this requires skill the same way manual coding requires skill, just a different kind of skill. Same with effort. And concentration. Engineering is just as hard as it has been. I am repeating myself. Code is cheap. Software isn’t.
But yes, if you are talking about “development/coding” skills, everything has been reset. You could’ve been a terrible developer pre-LLM and now be the best out there. Everything’s starting from scratch.
That seems short sighted and egoistic? Yeah, I know. Let’s face it —
The industry
There are mass layoffs. Companies reducing staff by 30%. People who used to have highly respected careers now desperately looking for a job. Millions of students from top universities struggling to find work in a market where there are enough senior developers already. And what took a 20-person team, one person can do now.
Yes. The industry is tough. Industrial Revolution level tough, maybe more. And mostly it is tough because of the high expectations a CS degree carries.
Computer Science is by far no longer the cheap entry, know how to code, get rich degree field anymore. And if by “Do not study Computer Science anymore” (remember the start of this text?) you meant exactly that, then yes, absolutely, I agree.
“It’s the hardest thing to get into today.” Probably true. It’s a hard industry, partly because of the state of the market, and partly because we are entering the field mid-transition. We are the generation that still learned to code manually, and yet might not write a single line of code later on. We are the ones getting in during the change, when nobody knows how jobs will look in five years.
“It’s the stupidest thing to get into.” I don’t think so.
What I mean to say
It is not about coding. Not at all. It never was.
It always was about solving problems. And now we’re solving different problems, faster than ever. The problem size is getting bigger the faster we (we, as in workforce with all tools available) can solve them. What used to be a 2 year project now is an afternoon. But this doesn’t mean there are no 2 year projects anymore.
Knowing how to code is important to know what’s going on in the world of LLMs and agents and understanding the foundational science is critical to develop LLM systems. And yes, scaling LLMs makes coding models better. LLMs get better at coding faster than anyone can learn to code, and I’ve said that years ago. But humans are not going to be competing with AI in writing code. Neither are humans competing with compilers in writing assembly.
The good news: Everything seems possible now. The fact that we feel like we would be able to build any application whatsoever, is an insane feeling. Software is democratized, they say. But there’s more to software than having it work. It’s about creating magical experiences, not a thing that “works”. Yes, coding no longer is a special ability. But neither is knowing stuff about taxes, business operations, sales, etc. It’s a shift, but the person who will always know most about taxes is the person who has been in the game long enough plus has intelligence at his fingertips. And the person who will always know best how to create something in tech will be the person who is deeply fascinated with how computers work.
CS is not the easy thing anymore, it’s the hardest in a sense. But if you are passionate, willing to show up, treat people with respect, work proactively and use all the tools out there, you are already ahead of most. Because there are simply so many people in the industry not out of passion, but because of money or status or the idea of what a CS career used to be. The industry rewards people who solve problems, think, build and ship. Not people with a piece of paper. And that analytical, deep way of thinking needs to be trained, which is exactly the role of a university degree. Remember, looking back, we are the ones who have shaped what’s next, and we are the ones who are shaping the future, today.
And again, this is my personal view. I’m a student out here, definitely not some industry professional. I wouldn’t have studied CS if the LLM revolution didn’t appear, and I couldn’t be happier that I did now. Different people have different views, and in the end you have to form and follow your own. What I am trying to show is perspective. A perspective that a lot of people are missing.
Thanks for reading to this point, and best of luck for whatever you are doing. Whether it’s studying for an exam, applying for an internship, a job, or chilling on the beach of Santa Theresa. You’ve shown that you care by reading this far, and I’d love to hear your thoughts.