I’m one of the best developers, and designers, and idea makers in this country. I could have made tons and tons of money doing what I love. But then I’m making these theories and projects that are not scientifically proven and not only I will not be making any money, but also will be crushed under the scrutiny. Why am I doing this to myself? Why can’t I stop?
God complex + inferiority complex + analysis paralysis + … = Pouya.
I no longer write isolated text. I add to the ARCHIVVM∙MAGNVM, like I add to my own mind. And I can trust that however my writing is, whatever problems it have, as much as I add to it, the massive body of ARCHIVVM∙MAGNVM, makes it possible for the LLMs to read it as a whole, and so the essays are now the true nodes of the Mind Graph, and I can have the LLMs read the whole and write the precise articles…
Mr. █████████ just called me and asked me to see them 9 in the morning at Sharif, I asked what happened to my ████████ training? He said, well that is gone…! Just come to your unit. I’m not going to ████████!
Foundation Models are still in their very early early stages. Much might happen to the AI world that we simply don’t know yet, and there is always the chance that nothing dramatic will follow. But until we have a verdict, there is another problem worth thinking about: what will happen to our jobs? I think I have finally found an answer.
The simple nature of an LLM is to predict the next token in a stream of tokens. The mathematics of prediction are straightforward: the more data you feed it, the higher the probability of the more frequently seen token being predicted. Simply put — if you show it a thousand pictures of a bird and ten pictures of a dog, which do you think it will draw better? Which do you think it will not draw at all?
A friend of mine was amazed at how agentic LLMs can write software. His most remarkable fascination was with how one had implemented a login page for his website — locking the user out after five wrong password attempts. And I agree. It is quite interesting.
But my breakthrough came when I realized: this is something we previously solved with frameworks and libraries. And there is a very interesting parallel here:
- LLMs are great at building login pages because every website has one. They have seen it thousands of times, so reproducing it is easy. But ask them to build something genuinely specific, something they likely haven’t encountered before, and they will struggle.
- Developers, on the other hand, never wrote login pages from scratch each time either. They considered it boilerplate — copied from a previous project or handled by a third-party framework. Their job existed for the specific things a project needed that no framework already solved.
What is happening now follows the same logic. These systems are solving the portion of a problem that was previously solved differently. When AI companies said they would help you write the boilerplate and free up time for the important work, they were mathematically correct.
And this is where it gets more interesting. Chain-of-thought techniques have advanced significantly, and agents are slowly taking over more and more of what people once did. But there is a ceiling — and it is a mathematical one. They do not go where the average won’t. An LLM, by its current architecture, cannot become a Maestro Da Vinci. It cannot be revolutionary, because revolution requires going against the statistical grain, doing what most people have not done and would not do.
I have been working with LLMs at their edge since they first became available. The week OpenAI opened their API, I had a fully automated newspaper running. An LLM would suggest topics. Another would choose and write about them. Two separate LLMs would independently verify each article, and only if both agreed the writing was sound would it be published. I had agents and chain-of-thought workflows before those were common terms.
And across all that time, one thing has remained consistent: these systems gravitate toward consensus. Toward what everyone agrees upon. Toward the norm.
So I believe truly creative work — doing what no one else is doing, thinking where the data is thin — has a better chance of surviving the AI transition than almost any other kind of work. The jobs most at risk are the ones where the market demands something common, repeatable, and expected.
When I was young, I would always make things that moved. My LEGO creations always had gears, electric motors, pneumatic jacks, and stuff like that in them. If I had made a castle, the doors were always opened with a gear system. It was my mind’s norm: things should be pretty, and have functions to entertain me. I never liked static things. I found them boring. The priority list was like this: Things that let you make other things are the interesting things—the more they let you the more they are interesting, things that move and have functions are fun, and other stuff, well they were to be dismissed. I loved LEGO, because not only it gave you the means to create other things, it had its own contained world. And I loved contained worlds. I loved Star Wars for example, George Lucas had made it like a LEGO: he had so many bricks (X-wing fighters, Millennium Falcon, Storm Troopers…), and each movie was a creation based on those bricks. The reason I got in love with programming was the same, Programming languages were like LEGOs, you could make infinity stuff with them, and they had bricks (if, for, variables, different colored grammars combining together just like the LEGO). And I never liked static things.
One day, I was around 8 or 9 years old, and I was looking at this amazing pot. There was some flowers in it, and it was beautiful. My Dad asked me what are you looking at? And I answered: “this pot”, and with a lots of shame I continued: “it is so beautiful, even though it does nothing.” his reply is still in me: “Not everything has to do something, somethings can just be pretty and there is nothing bad in it, you can make many beautiful things that don’t do anything”.
As years had passed, I tried, and today I have many things that don’t do that much, yet I still miss the old days of my life where it was saturated with things that did something, that were alive, and interesting. I had sighted for this state much of my life. When I had my hands on my first dumb phone, I had gone my way in the internet to find a super tiny spreadsheet, and something like a word processor. I wanted my tools to be tools, I hated when my peers just played with snake on them. On my first digital device, the second generation iPod Touch, I had everything that you can imagine, and not any games. Before they became mainstream like today, you could compose symphonies, write python/js/c#…, edit shaders, create modular synthesizers, … on my phone. It was and still is a powerhouse. I hate things that are just for consumption. There is no soul in them.
And I guess there is the same problem with science and the world right now. Everything is increasingly becoming more and more boring. There are many factors to this. One is the economics of late-stage-capitalism (R.I.P, and hail to our new lord neofeudalism.), which demands production in the maximum profit, minimum spending. If you are to create something today, your boss wishes you to outsource every single portion of it for faster and faster production. If you are a programmer, you would host in a cloud provider so that it is faster, use all the open source tools, and do everything you can to just use a framework and never write any code yourself. The result is pure boring production. The idea of the romantic person, doing things for passion, out of love and dedication… it’s dead. Today the creator has no good connection to the creation. This is multi-level alienation (Buddha bless Marx), and so, things are boring. On the other hand, science has much progress, but it is a dead medium. If you investigated scientific journals and their reports you would see an infinite stream of black and white pages, with charts, and nothing else in them. The only beauty you see in these papers is their remaining soul that Maestro Knuth made in TeX and is surviving in LaTeX. But that is it. When you have no poetry and soul in your writing, the science ends in no wisdom. It will always be: “new findings show that if you sleep early, you will have more longevity”, or in other words, finding facts, not wisdom.
Recently I’ve been reading that to Maestro Da Vinci the dream and the reality had perhaps no membrane. It was continuous. If he could realize them they would become creations, if he could prove them, they would become science, and if neither was an option, they either stayed at his notebooks or ended up in a fascinating theater. And that is a repeating theme in the greats. Einstein never started with the math. He had this fascinating question of what happens if two rays of light approach each other, how would they look like? He used to do the experiments in his minds (Gedankenexperiments as he would call them), and then find the math to explain them. Maestro Engelbart wanted to augment human intellect and he invented everything to get there, Maestro Kay wanted a paper in the hands of people and he did the same. That is why I love Maestro Victor. He makes revolutions without caring much about the scientific stuff of it, he creates, imagines, and dreams, and in between these there are worlds no one else has seen. Maestro Kay’s “view points” basically. I wish to be the same thing. I want to dream, to create, to go not for the glory of the science, but because my mind tells me too, and risk myself being a fool, make art of what is not true, and make science of what is.
I think we have reached a point where you either have to be a serious scientist, boring, and academic. Or you have to write science fiction, and never be taken seriously. I have read that MRI and Google Earth found their ideas from watching Star Trek. It is somewhat sad. Imagine how Isaac Newton wished to make gold with alchemy, imagine his crazy rituals and stuff, and how he invented calculus to explain natural philosophy. He didn’t know that his writing of PRINCIPIA was the Threshold of The Reverse Viewpoint from natural philosophy to science. And how much I miss that. Today’s science has no soul. Groups going to these sterilized conferences, getting funds through standard means, working with their profs to be the second or third on a paper, gathering academic scores… Where are people like Maestro Darwin? Where are those gentleman scientist who would go to crusades, be poetic and heroic about what they want to do, have so much beauty in their works and research? I know only of Maestro Victor, Maestro Kay, and Maestro Girba today who align to my visions.
I want a world with excitement. I want crazy, dreamy people who are making amazing things, and risking failure, not this safe guarded boring world where no one seems to be doing any risks, creating any new meaningful change. I love what I do, because even if my work proves to be all stupid, all wrong, all a waste, it will still be a lot crazy, a lot new, and a lot exciting. I’m not ashamed or afraid of doing my crusades, and doing them my way.