Coding in the Age of AI or AI Hype (not AI generated)

Coding in the Age of AI or AI Hype (not AI generated)

AI is a great help in coding. These days, discussions revolve around which chatbots people are we using, which coding agent is better, how many AI tools people know, how many AI subscriptions people are buying, how many coding agents they are using to write code, and how their jobs are easier or difficult because of claude code. The buzz is buzzing loud and clear. This raises the fundamental question that how people will learn coding, how one should approach learning a language, should we even bother learning as AI can generate the code and emphasis should be on learning how to use AI tools, what is the right mix where AI has a place and human judgement has a place. Do we just need to play with agents, or does coding has a place? all these questions are very interesting and worth pondering.

Do you remember how you learnt your first coding language, be it C, C++ (like me with salt and pepper hair). someone with more salt than pepper in their hair might think of languages like BASIC, COBOL. Someone with more pepper than salt in their hair can think of languages like kotlin, node.js, javascript. someone who has not had any salt so far can think of languages like python etc. regardless of your choice of language we all learnt basics. what are variables, how to assign values, what are primitive data types, how we pass values, how we divide the code into functions.

I remember that I must have read the 4 pillars of OOPS concept so many times. I would read it every year and every year that chapter sounded even more relevant. that book by herbert schild 'The complete reference to JAVA' I must have gone through it atleast 5-6 times. I remember After every read and coding for a few months it would sound more and more relatable and more and more important. There are few video series I refer over and over again, store them in your browser favourites for reviewing them over and over again. Every version of JAVA makes me wonder about the basics and makes me learn them even better. there were a few books I read to pass exams and certifications over and over again. Suddenly the discussions have changed. Up and until a few years back when chatgpt was not in the scene people were reading books or articles or watching video series for python to clear interview or have their basics in place.
Now all of the sudden question is not whether you know Java, scala or python. do you use AI? if so which model ? do you use chatbots Or agents? have you integrated them into your IDE? So the big million dollar question will be are those basic learning of language, cloud platform totally irrelevant or not important? Because the point seems to be do you know how to create agents in AI or not? 

I guess its worth taking a while to ask a few basic questions that every coder, budding coders, architects etc are pondering either consciously or subconsciously.

Do we need to learn coding and basics of coding ?

In my opinion, absolutely. In fact they are more than important now. AI will generate code for you, remove boiler plate code but you do need to understand the code well enough. In fact the importance of basics will become even more pronounced. If you are relying purely on AI to write code, you will make mistakes more and debugging will become even more difficult. Imagine a large problem (like an entire user story in an agile Sprint) is there and you gave AI and the AI gave you solution. You dont understand the code but you copy pasted the code and boom you think you are sorted. Thats not how its going to work. If that was true, we will not need more engineers going forward. It will be human who would break down a problem (A User story in a Agile sprint) into multiple parts. Use AI to code each part well. test each part well. think of the edge cases for each part and then think without AI and deliver the code. He/She should be able to read the code and understand it thoroughly. Make good comments while writing code so that others going forward understand how the code is working. Often I ask AI to write a piece of code. Many a times I find a complex lamdas are used in JAVA where I get confused. while the code is small I still prefer to be sure that I understand the code, I ask AI to break it down a little bit. I make a few changes myself and then test it. Use AI to help me design test cases and I write my own test cases. This ensures that if there are any issues in production or some breakage, it doesnt take much time to resolve and it can be either resolved easily and swiftly

How AI helps in learning and how AI helps in First Principles

Oftentimes whenever I start learning a new technology my first thought is why do I need to learn it in the first place. If you read a book or you watch video tutorials they are usually linear and have a quick path. When I started learning NoSQL, my first and burning question was why are you asking me to leave a technology that continous to work so well (SQL in this case). Often those burning questions are the ones that help you learn much, much faster than a book. Drilling deep into those questions allows you to absorb new technologies much faster and implement them for your projects or at the very least be prepared. I had that moment so many times. Learning JAVA 8 and its features. JAVA 8 was significantly different from the usual linear path of JAVA versions. When I am used to thinking in Classes and objects, suddenly I am thinking about functional programming and cutting classes. The ideas were not linear, and hence I had difficulty learning and absorbing. A video tutorial or book or web tutorial can not serve that purpose. AI makes it much faster. helps you absorb the concepts much faster and get a basic grip over the coding patterns. I had this when I started learning cloud. My biggest mental block was why now when on perm works just fine. Same thing happens when I start learning Queues or some messaging queue tech when REST APIs are just working fine. These burning questions and drilling them deep with AI chatbots allows me to learn more, comprehend more, absorb more, make better decisions, challenge my decisions etc. It was not possible as learning to code is by and large a lonely exercise and even lonelier when you are foraying into new tech all the time. Its like having a really intelligent friend who can answer as much as you want and as deep as you want. Allowing even a better experience than a genuinely smart human friend. 

Quantum of Learning

I remember the amount of code one would need to enter the IT world. With the amount of understanding today will not land you job. Its 2x 3x more learning before they start working. Everyone whether they have been there for 3 decades or just entering are constantly under pressure to learn more. Quantum of learning would be far more and hence use of AI would be inevitable and hence humans would be important because there are lot of things that have to be broken down and delivered.

AI will make us Nimble and agile as we have to learn, grow, deliver and unlearn all the time. While basics would be very very important we can let go of some semi manual tasks to AI and deliver more code, more features. People who can do the job of a developer, Analyst, Business and QA together will become valuable and everyone would be forced to be that way in one way shape or form. 

Misuse of Tokens

We know that AI is not free. There are huge infrastructure castles supporting them. They are extremely expensive. Just because we are use to using these chatbots for free or a very limited charge for premium subscriptions. If companies are provinding them with tokens then people are using them without giving a thought. A chatgpt or claude 8x to 12x more expensive. Slowly we will have to evaluate how to diligently use AI. The idea that use AI without thinking about resources used is not only environmentally taxing but also bad for writing good softwares. Already market is flooded with articles showing that people are using tokens 10x of their salaries begging the question is it really worth it from cost prespective. Its not cost but why this mania is not good. Dont use tokens unnecessarily. Use them wisely from the begining. People who lack basics will end up using more AI because they are compensating their lack of skills with AI

Conclusion

In simple terms your basics have to be strong and the basic knowhow has to be there to be functional. Assumption that we dont need to learn coding after watching some so called influencers is not only foolish but very risky. Just because AI can create agents solving your manual work is not replacement for humans who understand technology, languages, frameworks and design patterns. They can deliver far more but its very difficult to remove them from the scene.