The stated reason for renaming AI to SI is that Trump believes the word "artificial" suggests fakery and unreality, while today's AI capabilities have long outgrown what the term meant when coined — "super intelligence" is more accurate and carries more weight. But there are three deeper reasons. First, mobilization for tech competition with China: "whoever wins superintelligence wins everything" — SI carries the "win the future" narrative far better than AI. Second, the fight for global definitional power: whoever defines "super intelligence" dominates technology standards and discourse. Third, domestic political mobilization: paired with the October 8 "Science: A New Golden Age" $6 billion research plan, a new vocabulary was needed to rally consensus. Information as of 2026-10-13.
Trump's own words were blunt: "It's not artificial — we all agree on that."
The claim is not entirely baseless. "Artificial intelligence" was born at the 1956 Dartmouth conference, where "artificial" merely distinguished it from "natural intelligence" (the human brain). But in everyday English, artificial also means "man-made, fake, insincere". When AI mostly played chess and recognized images, the word was harmless; but when models began writing code, doing research, and passing the bar exam, "artificial" started to sound flippant, even pejorative.
The White House's logic: names shape perception, and perception shapes investment. A technology called "super intelligence" is easier for Congress, the public, and allies to take seriously than one called "artificial intelligence" — and easier to justify massive research spending for. This reason is real, but it is only the first layer.
On September 30, Trump showed his hand: "If we stop, somebody else will do it. Whoever wins superintelligence wins everything... Right now there are only two players, us and China."
Framing the rival as "us vs. China" and the contest's goal as "superintelligence" are a single move. AI is a broad technology cluster, hard to turn into a rallying cry; but "superintelligence" sounds like a definite finish line, something that can be "won" — like the moon landing. The rename repackaged a long technology race as a final with a finish line.
This also explains why the rename was immediately followed by the October 8 "Science: A New Golden Age" summit: $6 billion in investment, a $2.4 billion compute alliance, and the revival of national-level grand science missions. Without the "win superintelligence" narrative, that $6 billion would be just another research appropriation; with it, it becomes the opening act of a "new Manhattan Project".
This is the most important layer, according to critics — and the easiest to overlook.
In international technology competition, whoever defines the terms often also defines the standards, the ethics frameworks, and the market rules. The fight over standard-essential patents in the 5G era and over the definition of "advanced process" thresholds in semiconductors are precedents. If the official definition of "super intelligence" is first written into U.S. legal documents (the executive order already requires the White House science adviser to produce legislative text within about 60 days, roughly by the end of November), then when international bodies discuss AI governance in the future, the U.S. will arrive holding a ready-made definitional framework.
Even more interesting is the definitional technique itself: the executive order did not invent a new concept — it directly applied the existing U.S. legal definition of AI to SI. That means in American official discourse, "superintelligence" is no longer an academic concept about the future, but a legal instrument directly usable for procurement, regulation, and appropriations. This way of defining things means that when the government talks about "superintelligence", it is really talking about resources and power that can be deployed today.
| Perspective | Reading of the rename |
|---|---|
| The White House | A more accurate name, reflecting technical reality, rallying national investment consensus |
| Tech giants (responsive) | Aligning with the policy signal, competing for government contracts and partnerships |
| Tech giants (wait-and-see) | Terminology chaos raises communication costs; holding back for now |
| Legal scholars | Agencies changing wording is lawful; punishing private word choice would be unconstitutional |
| International observers | A bid for global technology definitional power and discursive dominance |
| AI safety community | Fears that generalizing "superintelligence" will dilute real alignment-risk discussion |
Don't overlook Trump's personal political style: he has a track record of using renaming as a political tool. In 2025 he renamed the "Gulf of Mexico" the "Gulf of America"; after the Associated Press refused to follow, the White House briefly restricted its reporters' access to press events, until a federal judge ordered it restored. For him, renaming was never just about language — it was a declaration of "who's in charge".
"SI" also serves a subtler political function: it drags the AI issue out of the "regulate vs. develop" framing. In recent years, American domestic discussion of AI had centered on risks, unemployment, and disinformation — broadly negative. According to public reporting, after a series of safety incidents, public sentiment toward AI kept declining. Swapping the name to "super intelligence" dresses the entire issue in a positive, forward-looking outfit — turning "should we restrain AI?" into "how do we win SI?" That rallies consensus at home and shapes the image abroad, killing two birds with one stone.
The objections cluster around three points. First, conceptual confusion: the AI industry uses "superintelligence" for AI that does not yet exist and surpasses humans; the government's equation of it with today's AI will make technical discussion and policy documents talk past each other. Second, constitutional risk: the October 8 "using AI makes you an enemy" remark was criticized as classic viewpoint discrimination — the moment it becomes action, it's a First Amendment case. Third, doubtful real-world effect: after the AI.gov revamp, the URL is still ai.gov, and the White House hasn't even finished renaming its own pages — if a "renaming campaign" can't unify the government internally, what can it unify externally?
Supporters counter: names genuinely shape public perception and willingness to invest, and America needs that mobilization in its race with China. Both sides are really arguing about the same thing — when AI becomes the core asset of great-power competition, even what it is called is no longer a technical question, but a question of power.
Not entirely. There is the formal September 29 executive order, the Justice Department's October 6 follow-up guidance, and the October 8 $6 billion research plan as a package — a coordinated combination. But the "enemies" remark has a strong improvisational flavor: there is currently no enforcement mechanism, and its legal force stops at federal agencies.
Directly, very little — China has its own terminology system and research plans. But the indirect effects deserve attention: if the U.S. writes its "super intelligence" definition into law and pushes it toward international standards, it will affect the discursive landscape of global AI governance, and China will need to respond at the standards level.
No. According to public reporting, the executive order does not touch private companies. Musk and Benioff's decisions to follow were commercial and political calculations; Altman's explicit refusal to follow had no consequences. Whether a company renames depends on how closely its business is tied to the government.
Yes. The terminology of academia and industry is not bound by a U.S. executive order. "AI" as a global standard term will not lose its status in the short term; what may actually change is the wording in U.S. government procurement and grant documents.
Behind the rename saga lies the full journey of the concept of "superintelligence" from academia to power. Read From AI to SI: this report systematically traces the academic origins of the superintelligence concept (from Bostrom to safety discussions at various labs), the technology road map from AI to ASI, and how great-power competition is reshaping technology discourse. Only by understanding the history of the concept can you see why a single renaming rattles nerves worldwide.
Read the full From AI to SI report