Learning Computer Science is still valuable

There is no denying that artificial intelligence has changed the way we write code. Almost every programmer I know, even some very old-school ones, use an AI code editor. And why not? It helps you write code faster, but not always better.

A few months ago on Quora, I responded to a question:

Former Nvidia CEO says coding is dead and AI will make English the new programming language. I was thinking of pursuing an undergraduate degree in CS. Should I still consider it or not?

My short answer was yes. Please feel free to see the full response Link to Quora topic

Why the Concern About Studying CS:

It really comes down to the job market. We see multiple articles or social media posts about AI replacing programmers or entry-level roles disappearing.

Bloomberg Article: Bloomberg about entry-level jobs becoming harder

New York Times: New York Times article about coders working at Chipotle

If the jobs are not there, why study computer science?

This is why:

At the end of the day, you will be talking to an artificial intelligence, which is a computer + some data + algorithms.

And ask yourself this: wouldn’t it be good to speak the same language as the computer?

Sure, the computer will understand human languages like English, Japanese, or Spanish, but any programmer will tell you that writing a command to a computer does not always give you the desired result.

Computer Science gives you the ability to understand why a computer is responding the way it does.

Ask yourself this: have you ever used ChatGPT or Gemini and got an answer that you knew was wrong or did not make sense? And do you know why, beyond saying it hallucinates?

How Does Talking to a Computer Work:

Computers take instructions written in a programming language and translate them into machine-readable code. Let’s take a look at two examples:

Assembly Language:

section .data

    hello db 'Hello, world!',0ah  ; Our string and a newline character

    len equ $ - hello             ; Length of our string

section .text

    global _start                 ; entry point for Linux

_start:

    ; write(STDOUT, hello, len)

    mov eax, 4                    ; sys_write system call number

    mov ebx, 1                    ; file descriptor 1 is stdout

    mov ecx, hello                ; address of string to output

    mov edx, len                  ; number of bytes

    int 0x80                      ; call kernel

    ; exit(0)

    mov eax, 1                    ; sys_exit system call number

    xor ebx, ebx                  ; exit code 0

    int 0x80                      ; call kernel

Python Language:

print("Hello, world!")

Now which one is easier: Python or Assembly?

Over time, computer languages became easier to communicate with, so we went from Assembly to Python. Both can do the same thing, but one is easier to read and write. And we are pretty much at the age where we can just say to any AI, “Print or Say Hello World.”

Get Francisco Quintana’s stories in your inbox

Join Medium for free to get updates from this writer.

Remember me for faster sign in

That being said, do you know what to do when you get this error:

-bash: python: command not found

Or this error in Assembly

-bash: nasm: command not found

Now your first thought might be why not ask the AI to debug. Sure that can work you can write “find python or nasm”.

But what if that still did not work? Wouldn’t it be nice to know why this error occurs and who you might need to speak to first? Hint: it will be your IT admin or whoever has root access to your system.

AI coding agents do not everything right:

Recently, a friend of mine was using one of these websites where an AI builds your website. Let’s call the website AIWebBuilder. The AI wrote the code for them, and then they wanted to connect it to the GPT type model. They wrote the prompts and “vibe coded” their way to a beautiful website. Then, when it came time to launch, on the AIWebBuilder’s “localhost,” it was not accessible to anyone but my friend.

So they asked AIWebBuilder to make it public, and they were charged to deploy it for them on top of the monthly subscription.

The site deployed, but it could not establish a connection to their GPT model. The problem was this GPT model was hosted on a private cloud, so when they tried to run it on AIWebBuilder, they got:

Service does not allow HTTP traffic ERROR

My friend used AIWebBuilder to debug, and they asked for an SSL certificate. And from here, they contacted me to help.

Their exact response was, “I do not understand why it is not working, I paid the subscription.”

Yet a developer knows what is going on, and within a couple of hours of work, the website worked. Then two weeks later, he asked me again for help.

What Computer Science teaches you:

It teaches you how to think like a computer and understand why the computer does what it does. Again, an AI system is just data + compute power + algorithms. Don’t you want to know how to speak to a computer?

Do you really understand what it means to upgrade to a GPU or what increasing “top K” means in your LLM?

Wouldn’t it be useful to know how mobile apps and websites work so you can solve problems faster?

With CS knowledge, you will understand what is going on, and this will help you build better products.

While most people will just prompt away, hoping the AI understands, you will have the knowledge to make it understand.

We are all working with AI, but most of us do not understand how to deploy, modify, or enhance an AI system.

Should you pursue a CS degree:

Being transparent, I did not study computer science at university because I did not have the prerequisites when I applied to university in the United Kingdom, and I also did not want to do a foundation year. But since 2015, I have worked in the software engineering field without a degree.

Do I wish that during my university years I studied CS?

The answer is yes, but I also feel it is not mandatory to study computer science to get a job. What I love about software engineering is that people care about what you build, not what your degree says. My first job in software, my boss told me he barely looked at my CV. He told me he looked at my GitHub projects and the websites I built, and that was good enough.

If you can afford the degree and plan to go to university, I say why not? If you cannot, I can recommend several courses, but my personal favorite is Harvard’s CS50: CS50 Online

But there are many you can choose from.

Whatever route you take, at the end of the day, we will all speak to computers, and knowing how they work will give you a huge advantage.