AI won't take your career

I keep on reading how AI will take my career every time a new model drops. On top of that, I see these ridiculous claims on how you don't need to read your code anymore, enterprise companies forcing AI everywhere, everyone adding a chatbot to their application and, to be direct, it's all BS to me.

Don't get me wrong, I love using AI, I use it every day! It truly changed the way I'm working. I use it to do code review, write code, even build out whole systems. This past month, I've been focusing on architecture work and it really is amazing what you can do with your AI tooling but...

You still have to read the code

Keyword here is tooling. Claims such as "you won't read code anymore" make no sense to me, and every engineer, even the ones building out the models, will agree that reading code is something that, currently, you can't take away from the whole process. AI has made us shift into reading code more than writing code, but in order to read code, you have to write code.

Anyone can clone anything now

In this day and age, it's so easy to build out a clone of anything. You can realistically take an application, tell Claude to build it and, assuming it makes no mistakes, you will have something that is decent. Hell, even get people to pay for it and to use it. And to make it even better it can last for some time which is amazing!

Weekend projects can become small MVPs and, to me, that is something that we couldn't really do in the past. There's a saying that software has become cheap, and I partly agree. I mean anyone can come and copy my project and call it a day.

The test of time

Where it all falls apart is one important thing: your architecture, and whether the thing you're building can pass the hardest test, the test of time.

• What will happen to your software in 2 years?
• How easy will it be to add a feature, or a new service?
• How easy will you be able to scale to 100, 1,000 or even a million users?
• Will your database be able to survive a million records?

Are you planning ahead, or are you letting your AI agent with limited knowledge handle that for you?

Who actually gets replaced

This has been said many times, but I'd like to repeat it, and I do believe people who write code blindly without checking it can get replaced. Hell, what's stopping me from having another Claude session which just writes code and doesn't check it? It's a lot cheaper to pay $200 than a whole salary. But that's not how things work. Unchecked code doesn't stay where you put it. It becomes what the next prompt builds on, and you won't find out it was wrong until you try to change something.

Learning how to learn

As a mentor, I always wanted people to learn how to learn. How to think about whether this code is going to be readable or even easy to extend in the future. My favorite question to people starting in IT is "If another person looks at this code, will they understand it?" The same applies to AI as well, will your favorite agent understand what is happening here? What are all the possible outcomes of your code?

Where I do let the agent in

And this matters even more when it comes to architecture. I wouldn't want any AI agent to build out my architecture without me seeing any of it. However, if you prompt your agent, go through multiple iterations and be a skeptic, then I would highly recommend using an AI for this, because it will help you do the "grunt work", the things nobody wants to do, like copy-pasting your code.

Even with all these things in mind, you still have to have rules which you want your coding agent to follow. Write them down, keep them in your head, do it the way you want, but for that to happen you HAVE to read your code. And to know what code is good and what code is bad you have to be a tinkerer and most importantly make mistakes.

Make mistakes on purpose

Don't be afraid to make mistakes, this is what makes you a better engineer or a better anything if you ask me. James Dyson went through 5,126 failed prototypes before the 5,127th one worked. And I'm not telling you to start your own business, but you have to know what's bad before knowing what's good.

To close off, don't be scared of AI and don't be afraid to use it and find what works for you. And most importantly don't forget to learn every day. Artificial Intelligence is an amazing tool and I won't stop using it anytime soon. But remember, the difference between a good and a bad engineer is how many times they've already been wrong about code they thought would hold up. And you can't buy that for $200 a month.