Introduction

September 29, 2026, saw some notable developments in the world of AI. One was the White House putting out a statement to instruct all government departments and agencies to use “Super Intelligence (SI)” instead of “Artificial Intelligence (AI).  In addition, the heads of most major AI platforms met at the White House to sign an accord on oversight controls and other measures, which included the same naming change. The fact sheet clarifies with the statement:

Super Intelligence, "conveys the true capabilities of the technologies being developed today," rather than "simply imitating or automating human intelligence as is implied by the term 'Artificial Intelligence.'"

I get the instinct. "AI" has a lot of baggage. For most people, the phrase summons HAL 9000, Skynet, and a robot with glowing red eyes, The Matrix, and that stigma makes it harder to talk about what this technology really does. A more accurate name would help. That’s kind of the problem, though: both terms already have definitions settled long before this week. So let's hold today's technology up against each one and see what the heck fits.

What AI Actually Means

The term "artificial intelligence" goes back to 1956, when John McCarthy and a group of researchers gathered at Dartmouth with a bold premise: that every aspect of learning, or any other feature of intelligence, could be described so precisely that a machine could be made to simulate it.

Note what that promises: not a machine that autocompletes a sentence, but a machine that reasons, learns from experience, understands what it's looking at, and carries what it learns into new situations. Researchers later called the full version "strong AI" or Artificial General Intelligence, to separate it from narrow tools that do one job well.

Ironically, the Hollywood version of AI, the computer that talks to you, understands you, and figures things out on its own, is much closer to the real definition than anything we have in 2026. The movies didn't get the term wrong. They got the destination right. We just haven't arrived.

Why Today's "AI" Doesn't Fit

Large language models are astonishing tools, and I use them every day. But under the hood, they're large statistical engines trained to predict what comes next in a sequence. They are spectacular at pattern matching across more text than any person could read in ten thousand lifetimes. That is not the same as understanding. These systems will happily write a flawless paragraph about a book that doesn't exist, with total confidence. They don't know what they know, and they don't know what they don't know.

They also don't learn the way the definition promises. Once training is finished, the model is frozen. It doesn't pick up a new skill from your conversation and carry it into tomorrow. Take it one step outside the territory it was trained on, and it can fail at a task a ten-year-old would handle.

So today's technology isn't the AI of the textbooks. The fact sheet is right about that much, but does the replacement fit any better?

What Super Intelligence Actually Means

Super Intelligence has a definition too, and it isn't "AI, but better."Philosopher Nick Bostrom popularized the term, describing it as an intellect that greatly exceeds human cognitive performance in virtually all domains of interest. Note, not just one domain. Not in chess, or protein folding, or writing code. Virtually all of them, by a wide margin.

So first you need general intelligence, a machine that matches a human mind across the board. You have to accomplish that first before you get something that surpasses the best of us at everything, including improving itself. We haven't reached the first step, let alone the top. As good as today's systems are, and with some new groundbreaking release on a weekly basis, it seems we still see them fall over on simple tasks at times, while blowing our minds with their ability to one-shot a first-person shooter game.

Here's what I find telling. The fact sheet doesn't define the term. It directs the president's science advisor to propose a federal definition at a later date. So the rename comes first, and the definition comes second. That's backward, and it leaves us in a bit of a pickle if it becomes a pissing contest between politicians; the Governor of California has already put out an executive order on September 30, 2026, insisting that AI remain the word. The reality, however, is that if Meta, X, NVIDIA, and Anthropic all decide to call it SI, then it’s going to be SI.

So What Should We Call It?

The goal behind the rename isn’t necessarily bad, and the complaint about the word "artificial" is fair, even if the cow has long since left the barn on this one. But swapping one inaccurate word for another doesn't fix that. And the new one carries its own Hollywood baggage. The all-powerful machine mind that outthinks humanity is the oldest sci-fi trope there is, which makes "Super Intelligence" an odd choice for a name meant to lower the temperature.

If we want accuracy, I'd rather see us name what this technology actually is: powerful machine learning that generates language, images, and code, and the early steps toward something that might someday earn the title of AI in the original sense. Then save "Super Intelligence" for the thing that meets Bostrom's definition, if it ever does. Machine Learning is really the most accurate pair of words, but it’s boring.

Better words are important. I just want ones that fit.