Where will I be when AI takes over

And here we are in 2026, and AI continues to haunt the plans of most IT professionals. I count myself among that group. As for being “haunted,” well, I believe I’m more of a skeptic than anything else.

Believe me, the amount of material I’ve consumed over the year trying to understand what’s happening is generous, to say the least. I empathize with everyone who has done the same; after all, who doesn’t want to know where this boat is going to anchor (or sink)? Since we can’t control the direction this is all heading, all we can do is read the situation as best as possible and act accordingly.

One thing is granted: AI is already impacting the way we work—whether through its own merit or the effect of a collective delusion combined with the non-stop propaganda from its owners. It’s here. It doesn’t matter if you’re a supporter or a hater. Being a hater makes it worse because, with 90% certainty, your boss has already asked how you’re using this “AI thing” to improve your work, even if it makes zero sense.

After nearly four years of daydreaming, I’ll venture to say that AI has consolidated itself as a productivity tool—and in some cases, a skill, the kind you put on a resume. While it’s vital for some professions, for others, the recommendation is to stay far away.

I will focus here on the viewpoint of someone using it as a supportive tool for their work. There are other important perspectives to consider, such as the consumer’s, but I’ll leave those for later or another text. The important thing is to try to understand the transformation and how to react to it.

IMPORTANT: This is my perspective on the subject as of the publication of this text. I do not intend to come back and update it if something changes. If that happens, I’ll likely just publish a new post.

The Effects of AI on Your Work

I’m a programmer—or a “software engineer” for those who find that term more professional. I like the latter, especially after hearing from a more experienced dev (I believe it was David Farley from Continuous Delivery) that “software engineering” is solving problems with the help of software, which sometimes involves coding. “Programmer” is just to remind us that, at the end of the day, the job is still about laying brick upon brick, or bit upon bit.

After many years of creating and maintaining software, I gave AI a chance to help me solve some problems. It was only a few experiments, but they allowed me to conclude the following:

It’s a Hell of a Personal Assistant

Focus on the word assistant. When iPhone’s Siri was launched, my mind nearly blew. Having something like JARVIS (from Iron Man) was far beyond anything done up to that point. Unfortunately, she doesn’t brew coffee, but answering questions with some context or executing simple tasks within her ecosystem was a massive leap.

Today’s chatbots are basically Siri with a brain 100x larger, maybe even more. But a word of caution: you don’t give Siri more than one command in a row, right? The next command comes with a new context because she won’t remember anything from the previous one. There is no chain of events or thought. Google Assistant tries and does a decent job, but it’s far from “complete.”

This, dear reader, gives you an idea of how far one of these chatbots can take you. Remember: it is an assistant, not an agent. The agent is still you.

There is Nothing Intelligent About It

Another programmer/influencer I really like (reminder to look up his name if I have time) says frequently and with authority: AI chatbots are nothing more than consumers and generators of tokens. When well-trained, they can generate well-structured results consistent with the problem to be solved. Remember, it only knows how to answer what has already been said to it, countless times, in various ways.

There is no intelligence in what it does, only aggregated “knowledge”—well-organized, yes, but the AI didn’t do that alone; it was “taught”, or directly given. There is no right or wrong. Whatever the model “knows,” it delivers. Period.

If you aren’t prepared to judge, analyze, and test what was generated, my recommendation is not to commit to the result.

Proof of this is the vast amount of stuff generated to promote content rather than actual products. So much so that announced products either fail or are taken down shortly after launch. Everything else is home projects that end up on GitHub to pad resumes. I’ve experienced both cases firsthand.

You Are Still the One Responsible

You know that Junior, you were put in charge of mentoring? Well, your AI assistant isn’t the same thing. It’s worse. The only similarities are:

  • You are responsible for everything your pupil does, whether it’s a junior or an AI.

That’s it. The similarities end there. Here is what your junior will do better than any AI:

  • Learn: AI doesn’t learn anything new (without the notorious, expensive, and time-consuming model training), not even via RAG.

  • Question: What a junior does best, IMHO, is challenge the status quo of many Seniors with new ideas and perspectives. A junior has no bad habits or bias, yet.

  • Evolve: If your hiring process is good and the junior is interested, you have a gem on your hands. You’ll not only have the satisfaction of watching them grow, but you’ll grow with them.

Here is what the AI will do better than your junior:

  • Quickly interpret error messages.

  • Answer questions rapidly (remember, it’s an assistant).

  • Always available (for a price), even during the odd hours some Seniors love to work.

  • Generate boilerplates.

  • Enable that POC for the idea you were too lazy to sit down and execute.

What doesn’t change is that no matter who does the work, you remain responsible. An AI speeds you up. A junior programmer transforms you.

But what if the Junior also uses AI? Well, we have a problem. If they are still juniors, how will they know how to prompt the AI correctly? How will they judge the quality of the result?

“But My AI Does Everything Automatically for Me”

If you’re one of those people who programmed your .md files to make Claude perfect or your N8N flows smoothly—first, congratulations; second, that’s not AI.

You “guided” your models via Markdown or pre-programmed your N8N steps to execute tasks under established conditions. There is NOTHING artificial or intelligent about that. It’s not artificial because you, a natural human (I hope), programmed it. It’s not intelligent because, if it were, it would have done it alone. This is pure deterministic programming, no more intelligent than a nested IF statement in an Excel cell.

If you want to call it AI to pad your LinkedIn profile, please do! Employers love it. Just acknowledge, for the sake of your future mental health, that this is about as much “AI” as the possibility of organic life existing on the surface of the sun.

What Changes?

Ignoring AI and its impacts is impossible. We are way past that point. The damage is done, and the effect on market mindsets is irreversible. You either embrace it, or you become legacy.

You can live without AI, but you’ll seem slower than others—in the same way someone who consulted a 2,000-page physical manual for a product seemed slower than someone who just Googled it.

The slower person earns less, is less important, less visible, and less relevant. If you choose to ignore it, prepare to be at the back of the line in many career races.

The best thing to do is learn the best ways to exploit what is already available—chatbots, prompts—to accelerate your work. In many companies, this is still very limited; in those cases, be cautious, but don’t isolate yourself.

You know that gap in relevance felt by a professional who masters Excel compared to someone who has no idea how it works? Swap “Excel” for “AI.” That’s basically it. Master the tool; don’t let it master you.

What Doesn’t Change?

The fundamentals. When I mentioned above that you need to master the tool, I meant knowing what it can do for you.

I love that word: tool. It can transform a human being into something powerful—a builder, tailor, musician, designer. But the overriding factor is you! If you don’t know how to use the tool, you’ll use a guitar to hammer a nail or a ballpoint pen to cook.

So, when you ask an AI to generate an app to do something, the quality will be analogous to your mastery of the subject—the fundamentals.

In the case of programming, my field, that would be, at a minimum: data structures, object orientation, algorithmic complexity, computer networks, relational algebra, etc. This, of course, added to the business domain knowledge of what you want to generate.

Don’t be that person who uses a violin to build a house.

Where I’ll Be When AI Takes Over Everything

Now that I’ve reached the end, I remembered the title of this article. What led me to write it was precisely thinking about what I can do best to face this entire transformation. I decided to dedicate 50% of my time to staying tuned to the innovations we need to absorb, while the other 50% will be spent on the basics—even if it means learning Assembly to understand how instructions are sent to a processor.

If the most likely scenario happens—that AI settles in merely as a luxury productivity tool—I’ll still be here doing the same thing, maybe even writing more texts like this (without AI).

But if AI takes over everything, well, I hope I’ve studied enough Assembly to program my own Game Boy games with all the free time I’ll have.


The author, who did not use AI to write this article.