Why Swapping Junior Devs for AI Is a Dumb Idea, and You Should Relax (for Now)

Many people ask my opinion about using generative AI to code. I'll be honest: I think it's amazing. If you know what you're doing, meaning you already have a solid programming foundation, AI will help you a lot with those countless tedious tasks that programming involves, like: writing tests (test scenarios, for example), boilerplate code, documentation, and so on.

But if you don't have that foundation, AI becomes a disastrous tool. It looks like everything will work out fine with a prompt and the push of a button, until things fall apart (and reality catches up with you).

Unsustainable architecture, basic security rules that aren't followed, data leaks… these are some of the things that start to happen.

And that's exactly why I want to convince you that if you're just starting your career in tech now, don't worry, you won't be replaced, at least not if you keep learning, studying, practicing, and using AI too.

So I'm going to lay out 3 reasons that sum up everything I think on the subject:

Reason 1: Some Tasks Disable Human Judgment

Doing everything I mentioned above with AI is really great, but leaving software architecture decisions, evaluating trade-offs, understanding business rules, modeling, and having a view of the software's past, present, and future are some of the tasks that require human judgment, experience, and intuition.

On the StackOverflow blog, in the article “Generative AI is not going to build your nursing team for you” ( link here ), it points out that “generated code often […] may not follow practices or conventions, may have fields that don't exist, made-up variables, etc.; it often requires detailed review, correction, and cohesion with the existing code”.

And yes, I'm the kind of guy who still likes using StackOverflow.

Reason 2: Exchanging Experience and Succession Planning

The only certainty we have in life is that one day we'll all “kick the bucket,” right? If that's true, it means that sooner or later, even senior-level professionals will eventually step away from their roles. When that happens, who's going to replace them? Robots?!

If companies replace juniors with AI, there will be a huge gap in the pipeline of people who would move into the next level (mid-level, senior), because no one will gain that foundational experience. This is pointed out by Matt Garman, CEO of AWS: he stated that “ replacing entry-level developers with AI is 'one of the dumbest things I've ever heard'” , precisely because companies need hands-on learning to build a strong team in the future.

Reason 3: We Love Junior Professionals

Every professional makes mistakes, especially the newer ones, but that's exactly why they learn so well. There are many reasons to love junior professionals.

First of all, their acquisition cost is infinitely cheaper than a senior's, and because of that companies invest in their development for a long-term return (that's how it was for me, for example ❤).

Besides that, they tend to learn very fast, have a fresh mindset and an impressive thirst for learning, which makes them great at adopting new technologies, being open to new ways of doing things, and bringing innovation to old practices (which often don't even make sense anymore).

Teams made up of different experience levels provide incredibly rich experiences for everyone, such as: mentoring, code reviews, brainstorming, and knowledge sharing.

It's important to remember that developing software isn't just about producing code. It's necessary to build a team, coherence, and scale all the know-how.

In Summary

Future seniority deficit : if there are no juniors, there will be no one to move up to higher levels; this can leave teams aging or with a shortage of people with mid-level or senior experience;

Product quality, stability, maintenance : AI can ( and will! ) introduce fragilities, obscure dependencies, and maintenance difficulties if whoever reviews it doesn't have deep knowledge.

Verification/adjustment overload : using AI isn't “press a button and it's all done”; a lot of junior work, even with AI, will still need to be reviewed, corrected, adjusted. If there's no one with experience, that cost can rise a lot or not be properly resolved

Matt Garman, the AWS CEO I mentioned earlier, said that replacing junior employees with AI is “one of the dumbest things I've ever heard”. And I couldn't agree more.

But then does AI pose no risk at all? Is my job guaranteed?

I didn't say that. I'd say that instead of thinking in terms of replacement, we need to think in terms of complementation. I believe developers use AI tools to speed up tasks and still need important knowledge and skills such as: understanding algorithms, critical thinking, problem-solving, communication, solid architecture fundamentals, security, maintenance, etc.