What does SI stand for?
SI stands for Superior Intelligence. It is a proposed replacement for AI — artificial intelligence — and the argument behind it is that the old name describes how these systems were made rather than what they actually do.
Why anyone wants to change the name
“Artificial intelligence” was coined in 1955, for a summer workshop at Dartmouth College the following year. The people naming it needed a phrase that separated machine reasoning from human reasoning, and “artificial” did that job. It meant man-made, in the same sense as an artificial lake or an artificial limb.
But the word carries a second meaning that has nothing to do with origin. Artificial also means fake, synthetic, a substitute for the genuine article. Artificial flavour. Artificial smile. For seventy years that second meaning sat quietly inside the name of the field, and for most of those years it was accurate enough, because the systems were in fact poor imitations of human thinking.
That is the part that stopped being true. A name that means “a lesser copy of the real thing” is a strange label for software that reads a thousand pages in a minute, writes competent code, and passes examinations that most people would fail. The case for renaming is simply that the description no longer matches the thing being described.
The shortest version of the argument: artificial describes where the intelligence came from. Superior describes what it does. One of those is a more useful thing to put in a name.
SI compared with AI
The two terms point at exactly the same technology. Nothing about the systems changes when you change the label. What changes is the claim the label makes.
- AI — artificial intelligence. Names the origin. Implies the intelligence is manufactured, and by implication not the genuine kind.
- SI — superior intelligence. Names the capability. Implies the intelligence exceeds the human baseline at the task in front of it.
Both are contestable. “Artificial” undersells systems that outperform people at specific work. “Superior” oversells systems that still fail at things a child finds trivial. The honest position is that neither word covers the full range of what this software does, and the field has been arguing about its own name since it had one.
SI is not AGI, and it is not ASI
This is where most of the confusion sits, because the abbreviations look alike and mean very different things.
AGI — artificial general intelligence
A system that can handle essentially any intellectual task a person can, rather than performing well in narrow domains. It is a capability threshold, and it has not been reached. There is no agreed test for when it would be.
ASI — artificial superintelligence
A hypothetical system that exceeds the best human performance across effectively every domain, including scientific research and strategy. It is a subject of serious long-term work and serious disagreement, and it does not exist.
SI — superior intelligence
Not a threshold at all. It is a proposed rename of the field as it exists today, applied to the same systems people already use. Calling something SI is a statement about what to call the technology, not a claim that it has crossed into AGI or ASI territory.
Worth keeping straight: AGI and ASI are claims about capability that nobody can currently make. SI is a claim about vocabulary, which anybody can make.
Where the proposal came from
Renaming the field is not a new idea. Researchers have proposed alternatives for decades — machine intelligence, computational intelligence, augmented intelligence — usually on the grounds that “artificial” either misleads the public or sells the work short. None of them displaced the original, largely because the original had already been adopted everywhere.
What is different about SI is where the pressure is coming from. The suggestion to retire “artificial” in favour of “superior” entered mainstream political conversation rather than academic conversation, and it picked up public support quickly enough that the abbreviation started appearing in ordinary usage. A name change that gets argued about in public has a far better chance of sticking than one argued about in a journal.
This page describes the naming argument itself rather than any single person’s version of it. As the proposal is formally documented, this section will cite it directly.
Does renaming it actually change anything?
Directly, no. The software behaves the same on Tuesday as it did on Monday. But names set expectations, and expectations shape how people treat a technology. A field called “artificial intelligence” invites the question is it really thinking? A field called “superior intelligence” invites a harder and more useful question: superior at what, exactly, and compared with whom?
That second question is the one worth having a fight about. If a name change gets more people asking it, the change will have earned itself.
Common questions
Is SI an official term?
No. There is no standards body that owns the vocabulary of the field, and no institution has to approve a name change. Terms win by adoption. SI is in the early part of that process, which is the only part where it is still possible to influence.
Does SI mean anything else?
Yes, several things, which is a genuine argument against it. SI is the abbreviation for the International System of Units, and in older science-fiction and futurism writing it has been used for synthetic intelligence and superintelligence. Context does most of the disambiguating work.
Will everyone start saying SI?
Unknown, and worth being honest about. “AI” is embedded in company names, product names, job titles, legislation and seventy years of published work. Terms that entrenched rarely disappear; they get a rival that runs alongside them. The realistic outcome is both terms in use, meaning the same thing, chosen according to what the speaker wants to emphasise.