The three phrases I hear most often in this period of AI explosion, from my fellow programmers, researchers and technicians, are these: “I am behind,” “I am not capable,” and “I feel useless.” As a New Year’s wish, I would like to tackle all three.
”I am behind”
This is a classic feeling. Why? Because there is always something new. Every week, every day, the best model in history comes out. There is a protocol war raging around artificial intelligence, with forty acronyms, and you do not know what the acronyms mean, what to bet on and what not to, and what the reality actually is. Obviously you feel like you are behind. The reality is that everything is extremely fragmented. There genuinely are not the fixed points you would expect, so maybe it is not you who is behind. It is just that it is all moving too fast, and nobody actually has the overview they pretend to have. They are all trying to carve out a bit of space, pretending they have understood something that cannot be understood, because it has not happened yet. It is too early to be sure of yourself.
So do not feel like you are behind. Nobody is really ahead on everything. These are all just explorations. To give you an example: if you go and look at Google’s A2A protocol, the agent-to-agent communication one, beautiful, it went around the world, well, the official implementations of that protocol were written by an intern, and they are written with their feet, because they spent more money launching the protocol than they spent taking care of the protocol itself. So maybe it is not us who are behind. Maybe it is them, playing it slick.
The other side of feeling behind is that the science underneath AI is objectively complicated, but the point is that you do not necessarily have to know it. How many people know how a SQL database works internally? They do not seem any less professional at using a SQL database. What you have to know is the surface of the commands you have available, the best practices, what you need to check, the risks. Artificial intelligence is the same thing. I spent years studying information theory, automatic differentiation, all stuff I now slap myself in the face over. You do not even need PyTorch or TensorFlow anymore, nobody builds networks by hand anymore, they have all become services, all APIs, so you do not have to become the math phenomenon, the neural network phenomenon. It is almost useless.
So, on the “I am behind” self-criticism, what is my invitation for this year? My invitation is to play. You are not behind, you are simply standing in front of something that changes constantly and is in full expansion. So the only thing you can do is play, explore. You do not have to stay on top of every piece, you have to try to understand what is happening and, above all, what crystallizes. It is impossible to know everything, you cannot be up to date. What you can do, though, is understand which of the things people are talking about actually stick over time, and even become boring. There is a lot of talk over time about AI novelties that then slide into the background and become boring, but keep recurring. Like MCP, like the concept of the agent, like tools. These things are becoming boring, and those are the ones that deserve your time.
”I am not capable”
This one, in my view, depends a lot on those LinkedIn posts too, people doing the demo, “look what I did with this library in fifteen lines of code, look what I made,” and the startup that does the super demo and raises forty million in a week. And the point is exactly this: they are demos. There is an exaggerated culture in AI of the demo, of the hype. Many of these things are nonsense, they are half-baked things, things where they spent months purely on the demo. In these cases the product they have, the one they are presenting, is not a real product. The demo is the product. So the fact that many people, in their private moment of self-criticism, question themselves with “I am not capable,” well, there too, maybe it is not that you are not capable. Maybe they are pulling your leg.
The truth, here too, is that many talk and few do. There is a tendency to oversell things that in substance are worth very little. And what we have to understand is that presenting a demo is very different from having mastery over something. The conversation tends to be black and white. Here too, even in the broader philosophical sense: it is intelligent, it is not intelligent, full automation, it does the job, no, it is useless. Always all or nothing, no shades in the online conversation about AI. And so you inherit this adolescent, black-and-white thinking where either you are capable or you are not, which is an equally idiotic idea.
In this case, what is my practical advice? The same as before: try to play in your spare moments, when you can, but with the pleasure of doing it, and understand what of everything that is happening I can bring into what I already know how to do. It is not that you have to become the AI expert, because these are tools, and there will be the pure AI expert, but that is a different job. For you, who maybe spent your life doing security and specialized in that, or you do front-end, or you do systems work, AI is about understanding what it offers for what you do best. How do I take these tools and use them for what I am good at? You do not have to become the new Geoffrey Hinton. This too, in my view, is a big misjudgment, thinking that AI is a world unto itself. You have your toolbox, you have your specialization: there is the wrench, the hammer, the screwdriver, and the pliers. These are tools. Artificial intelligence is one more thing in the toolbox. So the question is not “I am not capable,” the question is: of all these things I am seeing, what can I bring into my daily work? What could actually help me? And here too, always playing, without piling pressure on yourself, it is not true that you are not capable, it is that it is a mess. Nobody is truly capable in something moving this fast.
”I feel useless”
Let us move to the last self-criticism, the one I hear the most from my colleagues, the one I empathize with the most, and the one that gets to me every single time: “I feel useless.” This one, honestly, is the one that hurts most and the one to fight hardest, because in this period we feel called into question in our professionalism, our worth, our ability to bring something to society, to be useful, to put bread on the table. And here there is the shadow of vibe coding, right? These machines do everything by themselves, you do not need the programmer anymore, because the guy sits in front of Lovable and builds the whole application on his own.
And here I throw down the challenge. How come, if the machines do everything by themselves, then, lo and behold, the grief of actually building the thing and getting it online still lands on you? Could it be that you, the technician, end up being the one responsible for really landing something that is pitched at the social media and marketing level as the ultimate automation? But if it is the ultimate automation, why does part of the burden of figuring out how to use this stuff, and how to get it into production, come to you? And above all, a question I find to be totally toxic pressure: making this thing carry a certainty that it intrinsically does not have. How is it that in this whole AI world it has to be your responsibility for what happens and how things get implemented? Evidently, you are pretty useful.
Because what is missing? What is still missing is the verticalization of these things. This is software. It has to be taken, verticalized and made useful to someone, and it does not do that by itself, we have to do it. And that means knowing who this AI is being brought to, under what conditions, with what methodologies, with what technologies, and with the usual software cycle. When you get this kind of pressure, you can always say, “excuse me, wasn’t it all automatic, bro? So how is it that you saw the LinkedIn post, got yourself all hyped, and now I have to build the AI for you?”
Here too, what is the solution to this “I feel useless” pressure? It is not true that you are useless. On the contrary, your usefulness lies precisely in understanding how to take these things and make them actually usable, because on their own they absolutely are not. There is the concept of the human in the loop, there is the concept of AI system design, tied here too to what you know how to do, to your specialization. And it is all open, all still to be figured out. It is not that you are useless, it is that the usefulness of the technology is still to be found. But that is not on you.
Uncertainty is the sign of competence
To wrap up this whole thing, which is meant to be a bit of an encouragement: all this uncertainty you feel, that we feel, is actually the real sign of competence, in my view. You cannot help but feel disoriented in a world like this. You cannot help but perceive fragmentation and, in many cases, outright lies, especially from the ones investing billions in this stuff, who are terrified and so go around the conferences telling you “ah no, in six months programmers will not work anymore.” Go set yourself on fire, it is not true. So in all of this, feeling this insecurity is the sign of competence. It is a somewhat negative sensation, but it means you have perceived that they are pulling your leg. So let us see what of all this actually crystallizes, how to use it as a tool and not as a main specialization, because it is a tool. And third, understand the system designs that actually let people use these things safely, so that the responsibility does not fall entirely on us.
Making ourselves respected
To this I want to add a big point, again for my fellow technicians and researchers: we really do have to start making ourselves respected. As technicians we are used to being more in contact with machines, and I think this changes us. It changes us because we are constantly measuring ourselves against logic, and logic can crush you to the ground. It is one thing if someone comes to you and says “let’s do X.” That person is a human, so if you tell them no, I do not want to do X, they can invent a thousand pressures, manipulations and every other kind of nonsense to make you do X. But when it is your turn, and you are dealing with a machine, either you explain to it how to do the thing with logic, or it does not do it. And this, in my view, makes us super self-critical, because we are constantly crushed by logic. The programmer is a job that changes you, that sends you into a bit of background depression, because up to now, especially, there was no escape. When the computers do not work, you can curse and swear all you like, it is you who wrote the code badly. In most cases I think this leaves a mark.
But now AI changes even this a little, right? Because you can actually talk to these machines, so they too bring in this dose of uncertainty, this relational element, and in the near future I think we will find ourselves being a bit more like managers ourselves. Why is vibe coding interesting? Because the machines start writing part of our code, we keep an eye on them, we review, we explain how they should do it, but it is no longer strictly logical and strictly detailed like before. And this, in my view, can be an excellent opportunity for growth for anyone who has programmed their whole life.
The invitation is actually to make yourselves respected, that is, to recognize when your moment of insecurity really does come from some genuine gap of your own, in which case you self-criticize in order to grow, to change, and that is fine. But we tend to experience as a personal failing something we are not at all at fault for. Maybe it is not that you are missing something. Maybe it is the other guy who is an asshole, the one putting pressure on you, the one who saw the LinkedIn post, got all hyped, and even though it is AI I then flag the problem to him, you end up solving it anyway, or he goes around selling stuff that has nothing to do with the reality of what you are doing. These people need to be told to get lost. We have to respect ourselves by giving a smack on the head to whoever steps out of line.
To conclude, I do not know, it was meant to be an encouraging video and then it ended in headbutts, but my intent is absolutely one of affection, and it is to tell you: folks, we are not wrong, we are not behind, we are not useless. It is all still to be understood, really. Give yourselves time. Give yourselves a way to isolate a little from this flood of news and understand what is actually worth trying. Give yourselves time to explore ways of working with this AI that really is changing our job, but we have all the time, the desire and the competence to figure out how. It is not already decided, and you are not behind. Nobody knows. Happy New Year, folks, happy trip around the sun. Share this video with a colleague or friend who has impostor syndrome, so we can give them a boost of self-esteem.
