What AI Took From Me - And What It Gave Back • Bishal Roy

This is not going to be a regular blog that you see around about “how AI is coming for your job?”. In this blog, I am going to talk about my personal experience with AI.

I have always been a person who enjoys the process of writing code. I have spent countless hours making/learning the tools/technologies that made the process of learning code more enjoyable for me. Eventually, writing code by hand became a part of my identity.

When AI coding agents started coming into the picture, the sheer thought of it taking away the process of writing code by hand from me was scary. Initially AI agents were not strong enough to generate good quality of code. I used it as my coping mechanism to stay in denial that agents could write good code in the future.

In the last few months, we have seen a massive improvement in the ability of the agents to write code. And that left me in a weird spot as a software engineer who loves the process of writing code by hand. Seeing the industry relying on agents to generate code, I started to use them too. But I never found the satisfaction in generating code by agents. This experience pushed me into existential crisis.

The current scenario with AI

Today millions of line of codes are being generated by agents and then pushed to production. People stopped taking accountability of the code they are putting out. Relying on agents to the extent of outsourcing their ability to think is the reason, that there is no improvement in the quality of software. We are seeing more slop in the software industry more than ever. Open-source is one of the industries that is suffering because of it. It has become a lot easier to contribute to open-source projects. You don’t need to spend days or even weeks to understand the codebase before contributing. Just by writing few prompts you can understand the codebase and start contributing to it. So why are the open-source projects suffering when generating code has become so cheap and easier?

Yet on the flip side of the coin, there are open-source projects that are just work of art. This is because the people working on them are passionate and care a lot about the project. This is something which cannot be outsourced to agents. It is getting very important that you take ownership and care about the project you are working on. You should be able to explain how the project works. Why you used X instead of Y. so that you can at least correct and guide the agents towards writing good and maintainable code.

I still see a light at the end of the tunnel

It is clearly visible to me that in the coming years it won’t be viable to write code by hand (maybe most of it). The faster we embrace that, the better position we would be in future. After all writing code is only a small subset of software engineering, now we can focus on the other parts of this domain like system architecture and design, long-term codebase maintainability, technical trade-offs and so on. Deeply understanding how the system works and having ownership over the work we produce would be the most important trait we can have in this AI era. I still feel that when you are learning a new programming language or a new technology, writing the code by hand and struggling through it is the best way to learn. And if you are not clear with your fundamentals, it would be impossible for you to detect and correct the bad code that the agents will write. I also feel excited about the fact we can now build anything. The gap between what you know and what you don’t know has never been this small. And now I feel excited to see what comes up in the future. At the end of the day I am a software engineer, the tool I use doesn’t matter as long as I am creating good software.