The Iceberg and The Trampled Field

— A Story About Collective Amnesia

Imagine the collective mind as a field. For almost two years I have been trying to plant a seed in that field: the idea of an open, local AI, built to serve people instead of harvesting them. And what I observe, day after day, is something the news rarely points its camera at: giants fighting each other, trampling the soil back and forth, while the field itself (the public) closes in on itself, protecting whatever it can. Nothing gets absorbed anymore. Almost nothing answers. And the only thing still growing is what those giants breed inside their own private, walled gardens.

This is my attempt to turn the camera toward the crowd.

The iceberg

A few days ago I read a comment from a young person that I haven’t been able to shake off. Roughly: “AI has given abilities to those who already had wealth, and has taken wealth away from those who had acquired abilities.”

I understand the pain in that sentence. Something real is happening: the returns of this technology are concentrating, and plenty of people who did everything “right” are watching the ladder being pulled up behind them. But as a description of where technology comes from, the sentence gets it backwards. And that inversion shapes how whole societies react.

Ask around: who built the digital world? Most people answer with four corporate logos. Scratch the surface of that ice, carrying those logos, and you will find an iceberg.

If you use a Mac, the heart of your machine descends from NeXTSTEP: the Mach microkernel born at Carnegie Mellon and the BSD systems of the University of Berkeley. Academic work, public money. If you use Android, underneath everything beats Linux, a kernel originally written by a Finnish student in his bedroom. Python, the language in which most of today’s AI is written and trained, began as a Christmas side project by a researcher at a Dutch public research center. The Web itself was invented by a physicist at CERN and then given away. Git, which versions almost all the world’s software, was written in weeks by that same Finn, with no corporate mandate. And before all that, the personal computing revolution was argued into existence in hobbyist clubs and by kids typing assembly at 3 a.m. for the demoscene. I know, because I was one of them, in 1996.

Corporations did not invent this world. They arrived later, and did a magnificent job at packaging and distribution. The billions concentrated when distribution concentrated — not when the technology was invented. Confusing the tip of the iceberg with the iceberg is the amnesia that worries me — because a public that has forgotten who builds things will believe technology can only ever belong to billionaires. And that belief has consequences.

How the prophecy fulfills itself

Someone in my closest circle, a person who has successfully spent decades in front of huge audiences managing complex, mission-critical technology under enormous pressure, told me recently: “I want nothing to do with AI.” The revealing part is why. Not fear of the machines; the whole career proves the opposite. Fear of what is read and heard about them. If decades of hard-won technical confidence offer no immunity against the narrative alone, imagine the mind of a student finishing training right now, whose entire image of AI is second-hand: catastrophe headlines and CEOs announcing the end of work.

The numbers describe a generation bracing for impact. 47% of American college students have seriously considered changing their major because of AI; about one in six already did (Gallup–Lumina, 2026). Asked how AI’s impact on careers makes them feel, half answer “uncertain”, a third “concerned”, a third “anxious”, one in ten “depressed” — and only 7% “excited” (EAB, 2026). Among recent graduates, 89% fear AI will replace entry-level jobs (Monster, 2026). In Sweden, worry among under-35s doubled in a single year (Pew Global, 2026).

Here is the detail that exposes the machinery: fear tracks exposure in reverse. In the largest national fears survey, AI entered the top ten for the first time, but weekly users fear it less than those who never touch it, 46% vs 55% (Chapman, 2026). The fear comes from the story, not from the machine. The market quietly agrees: only 13% of employer job postings require AI skills, and workers who have them earn a 62% premium, up from 57% a year earlier (NACE; PwC, 2026).

So the loop closes. The public sees AI as billionaire property, refuses to touch it, the alternative ecosystem starves for users, and the prophecy fulfills itself: AI ends up residing only in the hands of those who kept building it. Refusing the technology does not make it disappear; it only guarantees that the only version that exists is the one built by the people you distrust. Abstention is a donation of the terrain.

And when resentment has nowhere to go, it turns physical, Waymo cars burned in Los Angeles, data-center protests across Ireland, the Netherlands, Aragón. The stage of “social incidents” is not coming. It has started, at small scale, the way everything starts.

The fear industry

Now add the broadcasters. The same industry that builds AI has spent two years selling its danger, and, remarkably, offering itself as the only possible custodian. In filings against the EU AI Act, the largest labs argued that the biggest developers are best positioned to meet the new requirements; translation: regulation as moat. In July 2026, over a thousand employees of the frontier labs signed a letter asking Washington for mechanisms to slow AI down; their employers endorsed it within forty-eight hours. Note the timing: one day earlier, a Chinese lab had published the open weights of a frontier-class model and was winning the headlines on price. In September, the CEO of Anthropic published an essay urging the industry to coordinate a slowdown; within a week, plaintiffs filed suit alleging that this coordination is, in effect, an illegal agreement among competitors. The same company’s IPO prospectus warns of “existential risks to humanity” while seeking a two-trillion-dollar valuation.

I admit, the fear occurs on top of true registered incidents, which is exactly why it works. And each round of this theater, whether it wins or loses in Congress (this one lost: the White House rejected the slowdowns), ends the same way: the legislative battle is forgotten, but the sediment of fear stays in the field, a little higher each time.

The silence

And this is the part nobody covers, the part I see in my own small communities every day. The public mind is not only afraid, it has nobody left to believe. Trust in the US federal government has fallen from 73% in 1958 to around 20% today (Pew). Trust in mass media hit its lowest recorded level after the pandemic (Gallup). COVID was the first global crisis lived through a pocket-sized channel of permanent contradiction, where experts, politicians and journalists visibly disagreed with themselves in real time. Whatever trust survived that does not survive a daily theater in which the builders of AI themselves alternately promise utopia, prophesy extinction, and lobby to be its only licensed wardens.

So the field does something any farmer would recognize: it stops absorbing. People don’t argue, don’t ask, don’t even criticize. Criticism would be engagement; engagement is risk. In a public mind trained by years of institutional betrayal and manufactured dread, even a small, free, beneficial project is processed as one more hazard to avoid. My own announcements meet not hostility… just silence. I used to read that silence as failure. Now I read it as the most honest symptom in this whole essay: an anesthetized public is not apathetic by nature. It was anesthetized.

Gardening, not fighting

You cannot outshout the giants; they own the loudspeakers. You cannot reason with a field. But gardens have always started the same way: small, local, tended by hand, in soil everyone else had written off.

The third way does not need to be invented. Capable open-weight models exist today. Local AI that sends nothing to anyone, that belongs to a family and not to a board, already runs on ordinary machines. Decades of experience as a programmer are not a requirement anymore to build amazing things: A non-programmer friend of mine, not an engineer, recently built an interactive classroom on top of our common AI integration framework: a live coding tutor that corrects and explains in real time, on his own hardware. And a poker table, because joy counts too. Nobody panicked while playing it. That is not a coincidence; it is the antidote working.

Ownership is the exit. Not resistance, not surrender: ownership. A public that remembers it can hold technology in its own hands stops being a field to trample and becomes soil again.

I’m just a regular guy who decided to start working non-stop, 22 months ago. I just love to see good ideas happening, and growing.

Rubén & Ainara