Artificial Intelligence has never been about reproducing human intelligence, and it's really great that most people don't understand this. It's not a gotcha, nor a hot take, it's a simple fact that has been true since the discipline was founded.
The men who coined the term were mathematicians, engineers, and programmers, and it is hard to picture they had any inkling of the havoc that the words "Artificial Intelligence" would wreak on the world, for generations.
It is equally unlikely they imagined the mythology AI would spark in the minds of men. The recent frenzy around LLMs is nothing new, just the grandest, latest accident in a decades-long beautiful, accidental scam.
For decades, Artificial Intelligence has lured all manner of talent with the seductive promise of its name. The things that humanity could accomplish with the help of intelligent machines... is not Artificial Intelligence at all.
AI is about formalized automation, about performing tasks autonomously without human labour. Yet, when applied at scale, it is so complex that it took some of Earth's brightest geniuses to figure out, and still it was just about counting, sorting and (gasp!) differentiating beans. Trillions of them. At a time.
The mechanisms that make humanity invest in what it romanticizes have for some reason decided they really dig ARTIFICIAL INTELLIGENCE. And the bounties have been many.
From search engines to airplane design, AI has been a constant, unseen companion in daily use for billions of human beings. Pathfinding, speech recognition, handwriting automation for postal sorting, collaborative filtering, machine learning for industrial optimization all came before generative language models.
AI has been capturing minds and purses long before ChatGPT was invented, and it was used to research the best airplane shapes, to sort through a planet's entire data to book the best restaurant in any city, to sort large image libraries, and to essentially manage the enormous amounts of data modern society is built on.
Artificial Intelligence is used daily by engineers and entrepreneurs building cancer detection systems, folding proteins to defeat crippling disease, or creating new crops to fight starvation. Every single one of them deserves every bounty of mankind's fascination for the dream of what Artificial Intelligence could mean, but definitely isn't.
However, with great power comes great proliferation of con men and zealots.
After mankind heavily invested in LLMs, when it reveled in the promise of a Jetsons future and asked "What else can we do?", they were told about Artificial General Intelligence.
They were told tales of a system that genuinely thinks, reasons, and understands the way a human being does. They were promised confident plans with timelines. They were presented with conferences, safety institutes and think tanks.
They were lied to by people who were either too dishonest or self-deluded to know that nothing in today's dominant AI trajectory leads that way. The founders of today's AI companies know they are telling stories, and it may even be that they half-believe them.
It is a difficult balance to strike, for those who understand the underlying technologies at play. There is a desire to inform the masses, to dispel the dream of artificial intelligence once and for all. It has been long enough, it has proven itself as a field that surely has yielded incredible advances decade after decade.
Surely, it will be enough to show that the mathematics need to continue to improve, that the tools will continue to deliver value and that the investments must continue to flow.
Certainly, the magic of AI will survive once everyone understands that LLMs are necromantic constructs that we use to speak with the long-dead spirits of every utterance mankind has ever breathed out, and no more?
Perhaps best to leave that for next week, and to let the grand scam continue just a little longer