Writing Code Was Never the Hard Part

it’s no secret that AI models are getting better at writing code. Since the release of ChatGPT, the progress has been incredible, and every few months they seem to become even more capable.

A lot of developers have tied their identity to writing code, and I understand why. It’s the most visible part of what we do.

But I want to offer a different perspective.

In my experience, writing code was never the hardest part of software engineering.

What Do I Mean by That?

Before AI, the hardest part of my job wasn’t typing code into an editor.

It was spending hours researching a problem, reading documentation, comparing Stack Overflow answers, testing different approaches, and figuring out why one solution was better than another.

Once I understood the solution and had proven it in a small test project, writing the actual code was usually the tiring part.

The next challenge was making sure that solution would scale. Could another developer understand it six months later? Was it easy to change? Was there a simpler approach? Was I solving the right problem in the first place?

Those questions have always been more difficult than writing the implementation.

If anything, writing code was often the repetitive part that many developers wished could be automated.

Should We Be Worried About LLMs and Job Security?

Personally, I don’t think we should panic.

Instead, we should appreciate that AI can remove some of the repetitive work and give us more time to focus on what really matters: understanding users, designing better systems, and building better products.

Being a software engineer has never been just about writing code quickly.

It’s about making good technical decisions, communicating with teammates, understanding trade-offs, debugging complex problems, thinking about maintainability, and taking responsibility for what you build.

Those responsibilities don’t disappear because an AI can generate code.

How I Think We Should Use LLMs

AI writing your code doesn’t mean you should skip the engineering process.

The process is still the valuable part.

Today, I think the ideal workflow is AI-assisted development, not blind vibe coding.

Think through your architecture first.

Choose technologies that actually fit your goals.

Have coding standards and patterns that you want your project to follow.

Then let AI generate an implementation.

After that, review it, refactor it, improve it, and repeat the process.

Think → Generate → Review → Refactor → Repeat.

You don’t always need the most powerful model either. Even smaller modern open-source models can be incredibly capable when you give them clear context and precise instructions.

Large frontier models certainly have advantages, but good engineering still matters more than model size.

Vibe coding often relies on vague prompts and hopes the AI gets everything right.

As developers, we shouldn’t aim to simply make something work.

We should aim to understand the solution, own the architecture, and make the code easy to change in the future.

Man vs. Machine?

It sounds like the plot of a movie: Man vs. Machine.

I don’t think that’s the right way to look at it.

The best use of AI isn’t replacing humans.

It’s enhancing humans.

AI should automate repetitive work so that people can spend more time thinking, designing, creating, and solving problems.

Think about the Mona Lisa.

Even if a robot could produce a visually identical painting, most people wouldn’t value it the same way as Leonardo da Vinci’s original.

Not because the robot lacked technical ability, but because the human work represents creativity, intention, emotion, countless decisions, and the story behind its creation.

Humans are valuable for much more than the output they produce.

Conclusion

Rather than fighting AI or letting it think for us, I think we should focus on integrating it into our workflow in a thoughtful way.

Let AI handle the repetitive parts.

Use the extra time to think more deeply, design better systems, understand users, and create better solutions.

That’s where real value has always come from.

The tools have changed.

The responsibility of building great software hasn’t.

💭 This is just my perspective.
Or maybe I've simply reached the final stage of grief: acceptance.

That's all for now.
you can follow me here on dev.to for more and on twitter @EmekaUgbanu

I’m building StepMello, a reflective walking app that helps you slow down, capture moments, and turn everyday walks into meaningful memories.
If that sounds interesting, I’d love your feedback.

👉 StepMello: https://stepmello.com