Skip to main content

Ai Reflections

· 3 min read

AI has consumed the industry’s attention like few topics before it. Somewhere along the way, we stopped asking the right question. Not “can this be built with AI?” but “what is the best way to build this?”

This space is where I explore AI from a different angle — not the hype, but the questions underneath it.

AI is a tool​

Every software engineer carries a toolbox containing multiple tools for solving problems. Clever software engineers regularly take care of their toolbox — updating outdated tools, removing the ones that are no longer relevant, and adding new ones. The whole point of carrying a toolbox and not a single tool is the fact that no single tool can solve all possible problems. This has been known since the very beginning of software engineering as a discipline. “There are no silver bullets,” as Fred Brooks once said.

Living with this truth, our job as software engineers is to take care of our toolbox and use the proper tool to solve the problem we are facing. And AI is just another tool — a powerful one, admittedly — that should be used with caution.

Even if it is hard to predict what will happen in our ever-changing industry, I think Fred Brooks’ thought will still be relevant. The ability to know when to use AI and when not to will differentiate clever software engineers from the rest. Do not let AI become your only hammer.

Use it or lose it​

Once we stop actively practising a skill, our ability to use it will deteriorate. This seems obvious and widely applicable. Let’s reformulate this generic claim to target coding: once we stop actively writing code, we will lose the ability to do it. The real question is what we stand to lose — what is the value of writing code itself?

I see writing code as similar to writing in general: it is a process of expressing thoughts. Several times during my career, I’ve been able to sense that something was wrong with the structure of the code, but I couldn’t clearly articulate the issue. I had to rewrite the code myself to uncover the actual code smell and understand how it could be improved.

For that reason, I don’t believe we are meant to be passive reviewers of generated code. The act of writing code is itself part of the thinking process.

I can’t express this idea better than Leslie Lamport did: “If you’re thinking without writing, you only think you’re thinking.”

Can AI produce intuitive code?​

We all know that a major task software engineers have is to reduce accidental complexity — to reduce complex structures and replace them with the opposite. But what is the opposite of complexity when it comes to software structures? Many would say it is simplicity, but I would argue that simplicity is not the goal. Intuitiveness is. The code should be intuitive to use, should not hide the intent, and should be easily understandable.

And here is the important question: can AI generate intuitive code? I will keep this one open. :)