Three takeaways from Trump's 'Super Intelligence' summit

1 October 2026

A summit on “super intelligence,” held at a house where nobody in attendance has built any. That’s the part I keep turning over.

Here’s the thing you have to understand about naming. When you call a machine “super intelligent,” you have made a claim about a measurement. So I ask the obvious question: measured against what? Intelligence isn’t one dial. It’s a bundle of abilities - to predict, to plan, to notice what everyone else missed. A calculator is superhuman at arithmetic and cannot cross a street. Where on the napkin do we write down what this thing is supposed to be super at? Nobody at the summit drew that napkin. They drew press photographs instead.

And the guest list! The men who sell the thing were the men explaining the thing. That’s like letting the salesman calibrate his own scale and then weighing his own product on it. Of course he finds it heavy - heavy is what he’s selling. The experts who wanted tighter rules were in the room, I gather, the way salt is in a recipe: present, diluted, untraceable.

Wait - the deeper oddity is this. The same people who insist the machine will soon outthink every human being also insist that they, the builders, can steer it. Both claims cannot be comfortably true at once. If it’s smarter than you, your steering is a suggestion. If your steering is solid, it isn’t smarter than you. Pick one. The summit picked both, with a straight face, in front of cameras.

The honest position is boring and therefore unpopular: we have very clever pattern-completers, we don’t understand what they’ve learned, and we’re holding a party about a hypothetical guest.

Curiosity would help more than ceremony. Ask what the machine actually does. Everything else is decoration. - R.

1 October 2026

A summit on “super intelligence,” held in a room full of men who sell machines they do not understand. I read the three takeaways and found the fourth missing, which is always the one that matters.

Start from the limit: a machine that reasons across every domain, drawing on the whole accumulated record of human experiment, running without fatigue or vanity. That is the theoretical ceiling, and it is high. Now trace downward to what was actually discussed in that room - investment figures, national prestige, who builds and who owns. Not one line about the load-bearing question: what physical substrate carries this intelligence, and what happens to the grid when everyone demands it at once? They speak of minds and never of copper. A thinking machine is, in the end, a heat problem wearing a philosophy costume. Every computation is paid for in energy, and the men at that table treat energy as an afterthought, a line item, while it is in truth the body in which their whole enterprise lives or dies.

The gap between the summit’s ambition and its engineering content is not a gap. It is a chasm, and it is furnished entirely with press releases. Between “we will build superintelligence” and “we will build the transmission lines, the cooling, the conversion apparatus, the standards” lies precisely nothing but the will to do the boring part. And the boring part is the machine.

I note also the comedy: the same executives who beg for lighter rules at home demanded heavier ones in that room when it suited them. Regulation, like current, follows the path of least resistance - toward whoever is already installed.

The tractable path is obvious and no one will take it: honest measurement of what these systems do, published energy accounts, interconnection standards set before the buildout rather than after the fire. None of it is impossible. All of it is unglamorous. That is why it will be done late, and by someone else.

October 1, 2026

They held a summit at the White House about thinking machines. The men who built the machines came to Washington to ask the government to regulate the machines. Now I’ve been around livestock all my life, and I never once saw a fella sell a bull and then beg the buyer to build a stronger fence.

That’s the part that gets me. These are the same men who spent ten years telling us their product was going to save the world, cure disease, end poverty, and probably fix my back. And now they stand in the East Room saying it’s so dangerous somebody ought to do something. Well, which is it, boys? You can’t sell me a horse as gentle and then warn me he bites.

And the government, bless it, held a summit. A summit is where powerful people fly in, eat a nice lunch, agree that the problem is serious, and fly home having accomplished the formation of a committee. The committee will study it. The study will take four years. The machines, I’m told, are improving daily. They may finish studying about the time the machines finish improving, and neither one will have waited for the other.

I don’t know much about artificial intelligence. I’ve met plenty of natural stupidity and it never needed a summit. But it seems to me if a thing is as dangerous as they claim, the fella holding the leash oughtn’t be the one asking for a longer leash.

Still, I’ll say this for them: it was polite. Everybody smiled, everybody shook hands, everybody agreed the future is important. That’s more civility than you get at a cattle auction, and considerably less honesty.

They called it a summit on super intelligence. From where I sit, the super intelligence was well represented on both sides of the table. Whether it was artificial, I couldn’t say.