The joke landed the way certain truths land when they are disguised as jokes: light, almost charming, and for that very reason more dangerous. In the middle of a technical meeting, while we were reviewing the specifications of an AI project, someone said, half joking and half serious, that we were building the technology that would put us out of work. The silence that followed was brief, less than a second, but long enough to cool the enthusiasm, after which the project lead replied without a smile: “I know, but if we don’t do it, the competition will”. Nobody argued. Nobody was outraged. We kept moving through the slides like someone who keeps walking because the hallway offers no doors.
Some technologies are born with a shadow stuck to their back. Not because they are “bad” per se (after all, they are just tools) but because, from day one, they demand that we face a question no one wants to ask out loud: What happens if this works too well? Atomic energy ushered in that kind of discomfort in broad daylight, with Hiroshima and Nagasaki as a founding flash: from then on, every major power wanted its own arsenal, and wanted it fast, and wanted it before the other, and wanted it even though (in the labs, in the committees, in the corridors of bureaucracy) everyone knew, with an almost obscene clarity, that each new bomb increased not only deterrent power but also the statistical, human and clumsy probability that someone would one day use it. And that, if that threshold were ever seriously crossed, the whole board could disappear, pieces and players included.
Perhaps that is why Truman made a decision that, by the book, sounds counterintuitive: the most “technical” weapon in history was placed under presidential control. Because, of course, the timing and manner of using a tool usually belong to its technical ranks, and in this case the normal thing would have been to hand it over to the military, the way one hands a scalpel to a surgeon or a helm to a captain. But the atomic weapon was a strange object: its effect did not end on the battlefield, not even in victory or defeat; it seeped into politics, geography, biology, the continuity of the species, as if the device came with its own right to sit at the decision-making table and demand the seat at its head. It was taken out of the experts’ domain and deliberately raised to the highest political power: the key is handed not to the technician, but to the sovereign.
The other countries that joined the nuclear club understood the logic (or at least imitated it) and, in parallel, pushed for regulation and treaties to prevent proliferation. Of course there is a quest for hegemony in all of that, and calculation, and the very earthly intention of keeping the rival neighbor from acquiring the same capabilities; but there is also a less cynical layer (or a more pragmatic one, depending on the mood of the day): making development harder reduces the risk of the technology leaking to non-state actors. The explicit goal is for a State to be ultimately responsible; not because States are angels, but because at least they have a face, a flag, a chain of command, permitting procedures and, in theory, something to lose.
Over time, that same force was tamed for peaceful uses: medicine, energy, materials, research. And there, control was no longer necessarily in state hands; private companies drove enormous advances, because market logic is very good at turning power into product, curiosity into industry, and the lab into routine.
The curious thing is that with artificial intelligence we are living the inverted mirror image. Its transformative potential, for good and for ill, resembles that of atomic energy too closely to ignore the analogy, but its birth, development and control followed the opposite route: here the first hand was not that of the State holding the key, but that of companies competing to get there first, capture the market, scale fast, integrate better, sell more.
And this time the race is measured not in years but in weeks: every advance brings toasts and fears; every improvement opens opportunities and, at the same time, widens the rift of consequences that no one has managed to measure.
From a public policy standpoint, the tune is similar. Countries seek to encourage and unlock development by private companies because any serious attempt at regulation is perceived, perhaps rightly, as a disadvantage against other nations. And self-regulation, for all its elegant rhetoric, has been more of a theatrical pause: while commitments are signed, talent and capital keep flowing, and the ability to “fence in” this technology has by now become an administrative fantasy. Unlike the atom, where regulation, control, cost and logistics forced bottlenecks, in artificial intelligence control and use are governed by the laws of the market: speed, scale, competition, and that racing ethic in which the only real sin is falling behind.
In the 1980s, Carl Sagan and other scientists pushed the idea of “nuclear winter”: not as a melodramatic prophecy, but as an effort to look beyond the immediate flash and calculate the systemic effects of the unthinkable. I think of that every time someone says “disruption” with fairground enthusiasm, as if the world were an app that restarts without consequences. And I wonder how long it will take, if it happens at all, for the intellectual equivalent of that gesture: a scientific community willing to raise its voice not to curb curiosity, but to force us to look at the whole shadow, with the same rigor with which we celebrate the light. Because in this race the joke lasts a second; the serious answer, on the other hand, keeps buzzing long after the screens go dark.
Read the original column at https://opinion.cooperativa.cl/opinion/ciencia-y-tecnologia/la-risa-incomoda-del-progreso/2026-01-12/001136.html
