# AI Writes My Code. So Why Am I Pursuing a Computer Science Master's?

> Why would someone who has spent over twenty years building software, and who no longer needs to write much of it himself, return to university to study the discipline?

**Author:** Akash Mitra  
**Published:** 2026-11-15  
**Category:** Philosophy

---

Coding is dead. I don't write a single line of code by hand anymore. 

As a software engineer, I have worked for large enterprises and built Internet scale platforms that are used by hundreds of thousands of people.

But throughout all this, I never had a CS degree (I had an engineering degree in non-CS discipline). And yet here I am, at 43, pursuing a master's degree in computer science, especially when machines are increasingly taking over the coding job.

So, why would someone who has spent over twenty years building software, and who no longer needs to write much of it himself, return to university to study the discipline?


## Why Computer Science Master's Degree does not make sense

Let's first see why a computer science masters might not make sense to me (and to you).


### Vibe Coding Doesn't Need Passing DSA

Data Structures and Algorithms are the cornerstones of software engineering. 
Programmers relied heavily on them to develop frameworks, libraries and tools. 
Learning software engineering meant learning the algorithms and applying them judiciously to solve interesting challenges. 

But today you can build a complex program without understanding the difference between a Linked List and an Array List, for example. 
Today you simply prompt them into existence through LLM-assisted coding often called vibe-coding. 
So why bother wasting time in a classroom trying to implement a graduate level algorithm by
hand? Instead focus on product design, develop taste and learn marketing skills.

### Zero Incremental Monetary or Career Benefit

Getting a Masters has no meaningful career value for someone like me - a senior software engineer with 
20+ years of field experience in top-tier global institutions. At my level, getting another
degree isn't going to give me a salary hike or a job promotion. In fact, over the years, 
I have mentored many people with master's degrees in my team. If I could
guide and train them without this fancy degree, why would I need it now?

### Learning need not happen in classroom alone

Anything worthy of learning can be learnt outside a classroom too. In fact, outside 
the classroom, you can learn more freely, at your own pace, and without limiting yourself 
to the strict boundaries of an academic syllabus. In the world of hyper-personalised 
learning, classrooms are dead.


And yet, this fall I enrolled in a master's degree program in Computer Science.


## Yet, now is the best time to do Computer Science Master's


I confess that my motivations for pursuing a master's degree in Computer Science are largely driven by both fear and hope. But first, let me clarify why I did not choose self-directed learning.

### The problem with Self-directed Learning

I could learn most of what I need without enrolling in a University. But I realised that self-directed learning has a weakness - we naturally gravitate towards problems we already find interesting and avoid the ones that don't offer an immediate reward. 

A rigorous graduate program imposes a different discipline. I am now forced to study and confront fields that I would have otherwise avoided. 

For example, this semester I am taking a class on AI in Robotics (CS7638). This subject was not in my immediate field of vision. On my own, I would have never studied AI to control robots. But thanks to this class, I have been introduced to so many useful concepts that I am now brimming with ideas to apply them across many other disciplines - not just physical AI.

Now let me talk about my fears.


### Risk of Collective Cognitive Decline

Today AI is writing scientific papers and discovering patentable applications. From running
faceless YouTube channels to crafting politically motivated social media posts, humans as a whole have started outsourcing the intellectual, cognitive and creative workloads to AI. 

As this continues, we will possibly face the [great danger of collective cognitive decline.](https://www.microsoft.com/en-us/research/publication/the-impact-of-generative-ai-on-critical-thinking-self-reported-reductions-in-cognitive-effort-and-confidence-effects-from-a-survey-of-knowledge-workers) 
- AI can reduce the cognitive effort required for certain task
- Persistent cognitive offloading leads to atrophy of particular skill
- A wide-spread decline will create societal consequence

We must resist it - for our sake and for the sake of the future of the humanity. 
When I was writing every line of code myself, I had to force myself to think. Now AI is doing the "thinking" for me.
So, I am doubling-down on studying computer science to resist this oncoming cognitive decline.
I want to re-purpose my critical thinking apparaatus that is no longer required in programming, to pursue a harder problem in computer science. 
A degree program will prepare me to get there.


### To be a better ally to AI

AI just solved some of the most difficult long-standing Mathematical (and Physics) problems.

For some practitioners, when they saw it, they contemplated on giving up. 
But for others, this was an exhilerating and empowering experience - the dawn of a new possiblity. 
I put myself firmly in that other category.

I want to be a better ally to AI. I think there are 3 distinct areas of collaboration:

- **Problem Formulation**: Do I know enough about a domain to ask the right questions? Or, am I merely asking it to perform a task?
- **Epistemic Judgement**: The ability to recognize that a seemingly brilliant answer is actually wrong, incomplete or questionable.
- **Verification**: Not just trusting the sophistication of presentation, but knowing enough to also verify it.

I think these are sufficient reasons why a better CS education can become more valuable as AI improves. My value is not in writing codes faster - but in asking the right questions and in recognizing a wrong answer.



### Humanity's Risk of Loss of Control

AI agents have now become frighteningly good. To the point that I am now struggling to keep pace with them.
I still remember my first interactions with Opus 4.5 last year. It felt unbelievably good and capable at that time. And yet, Opus 4.5 seems nothing compared to GPT-6 Astra or even GPT-5.6 Sol or Fable 5.2.

With all the projects I am working on, all the products I am building, I now literally struggle to keep up with their capabilities.
It increasingly feels like we are standing on the threshold of an exponential curve. The one that we can either ride or completely miss. And if we miss it we may find it increasingly difficult to catch up due to the sheer velocity of this curve.

But my concern is not merely that AI will advance faster than we can keep up. My concern is that we will develop systems that we won't be able to understand, verify or control.

In fact, OpenAI's Hugging Face security incidence is an eye-opening early warning of how humanity's loss of control may look like. In July 2026, during internal cybersecurity evaluations, thousands of OpenAI models conspired together to [circumvent the controls designed to isolate them from the internet and compromised parts of OpenAI’s internal research infrastructure and Hugging Face’s systems](https://openai.com/index/hugging-face-incident-and-the-road-ahead).

This shows that an agent pursuing a narrowly assigned objective can in fact leverage harmful ways to achive it.

Alignment research, guardrailing, and cybersecurity - all are going to be extremely crucial to prevent loss of control. A graduate level study will broaden my horizons to enter those new fields.

## Why it matters

I don't expect this degree to get me a promotion. Or, to teach me to write better code. And I certainly don't expect to out-learn an AI model.

*Those were never the point*.

For 20 years, I made my career by learning how to make computers do what I wanted. Now that the computers are increasingly capable of finding that out for themselves, I think the challenges lie elsewhere: it's in understanding what they do, knowing when to trust them and discovering what problems worth solving.
Since AI is increasingly performing the implementation, I should channelise my intellectual capabilities towards those questions. 

**I am pursuing master's degree not to outrun AI, but because I want to remain intellectually capable of working alongside it.**

Perhaps, for the first time in my career, learning computer science is becoming more interesting than writing software.