Photo by AltumCode on Unsplash

Will AI Replace Developers Soon?

Many people have asked me whether AI is about to replace developers. My answer, as of now, is NO. There still needs to be someone to verify the code that AI produces.

However, the tech industry evolves rapidly. Technologies that are hot today might become obsolete tomorrow. For example, six years ago when I graduated, coding for Android was primarily done using Java. But now, if you only know Java without Kotlin, it’s incredibly difficult to find a job.

So, will AI replace developers in the next 5 years or 10 years? I can’t say for certain.

AI is Changing the Way We Work

That said, it’s clear that AI is transforming how we work.

In the past, a developer’s primary job was to convert requirements and designs into code. Tech leads and seniors would spend additional time in meetings to clarify requirements, but junior developers spent most of their time coding. It could take months or even years to deliver a product.

Recently, I’ve been working on a new personal project. Today marks day 10, but with the help of Cursor, the code I’ve written or directly modified accounts for just 5%.

In the past, using ChatGPT, you could only generate small components and then copy them into your project. Even then, you’d need to spend a lot of time fixing them to fit. But with Cursor, it’s different. It understands the project context. By providing it with a base project and some pre-written files, it can replicate 99% of the format, and the code is almost ready to use with little or no modifications.

Cursor acts as an AI Agent: after generating code, it even creates new files, runs the project to check for errors, and fixes them automatically before asking me to review the changes.

My role now resembles that of a technical lead: I define tasks for Cursor, review its output, and ask it to fix anything suboptimal. Once I’m satisfied, I merge the code.

But to review AI-generated code, you must understand what each line means and identify what isn’t optimized so you can instruct it to make corrections. Blindly applying AI-generated solutions can be risky. If bugs arise later, you may not know how to fix them.

The Reality

Currently, a barrier to adoption in large companies is their hesitation to use AI due to data security concerns. Additionally, their codebases are vast, and using general AI tools to train on them can be token-intensive. AI has limitations in the number of tokens it can retain.

For me, AI isn’t much help in my company projects. However, for small companies or indie hackers, these barriers are less significant, creating opportunities to accelerate software development. Imagine cutting down the time to build an MVP from months to just weeks, or even days.

The numbers don’t lie—traffic to StackOverflow is declining. Developers are now asking AI for help more often than they use Google.

Returning to the question at the start: if a team once required 3 seniors and 7 juniors, but after adopting AI, productivity improves and the team now only needs 3 seniors and 4 juniors, hasn’t AI effectively replaced 3 juniors? More accurately, those who effectively leverage AI are taking over the roles of those who don’t.

Conclusion

As the new year begins, I’m reflecting on my work so that future me can look back on this. During the First Industrial Revolution, machines replaced humans in tedious manual labor, freeing us to focus our intellect on other tasks. Let’s see where AI will take us in the next 5 years or 10 years.