Names Are Governance

· 5 min read policygovernanceterminology

Navy graphic with amber headline text reading Names Are Governance.

On Tuesday, President Trump signed an executive order titled “Inaugurating the Era of Super Intelligence.” The order directs the executive branch to replace “artificial intelligence” and “AI” with “Super Intelligence” and “SI” in official correspondence, websites, reports, and policy documents, according to The Register’s coverage. It is not retroactive, and it does not change the underlying statutory definition: HSToday’s reporting notes that “SI” initially covers exactly the same systems as the federal government’s current legal definition of AI. The director of the Office of Science and Technology Policy has 60 days to propose a formal federal definition and say whether it should modify or replace the existing one.

At the America.gov launch that morning, Trump previewed the rationale: “We don’t want to hear artificial because it’s not artificial. It’s very powerful,” HSToday reports. He had been test-driving the term since his UN General Assembly address a week earlier.

The objection sounds odd until you parse it. “Artificial” names origin. The systems are built, not born. That is a fact about where they came from. “Super” names a capability claim, and it is measured against a human baseline: in the field’s own literature, a superintelligence is an intellect that greatly exceeds the cognitive performance of humans in virtually all domains of interest, per Wikipedia’s summary of Bostrom’s definition. Origin and capability are orthogonal predicates. You can have artificial things that are not super, and super things that are not artificial. Swapping the first word for the second commits a category error, and the order itself proves it: the label changed while the statutory definition underneath it stayed exactly the same. If the rename were about precision, the definition would have moved too.

It is worse than imprecision. The word already has a job. In AI research, “superintelligence” names something that does not exist yet: intellect far beyond the best human minds, which is why the safety literature is written about it rather than for it. Applying that word to every classifier, chatbot, and recommender in the federal inventory smuggles an unproven capability claim into official vocabulary. From now on, every federal document that mentions the technology will implicitly assert that the systems it regulates greatly exceed human cognition. That is a capability claim dressed up as a noun, and capability claims belong in evaluations and evidence, not in names.

There is a reason for the framing, and it is not precision either. The rename presents the technology as triumph rather than risk. Vocabulary that presumes triumph makes caution read as mismatched: it is hard to sound the alarm about “super intelligence” without sounding like you missed the memo. I am not arguing motive; I am describing mechanics. Words set the frame, and the frame sets which arguments are legible. At the UN a week earlier, Trump dismissed pressure for tougher industry regulation; the renaming and the regulatory posture are one message in two media.

The irony is that the field’s founding name was itself a branding move. John McCarthy coined “artificial intelligence” in the 1955 Dartmouth funding proposal, and in his own later telling, “I invented the term artificial intelligence when we were trying to get money for a summer study,” as Engadget’s history recounts. He chose it partly to escape association with “cybernetics” and Norbert Wiener’s shadow, per the Dartmouth workshop history. So the first name was chosen for funding appeal and field positioning. Seventy years later, a second rebranding arrives by executive order, again for positioning. The difference is that McCarthy’s name made no capability claim. It just said where the things came from.

I have a counter-proposal, one I made this week in the same conversation where I rejected the federal rename: the honest term is “synthetic intelligence.” It names origin, like “artificial,” without the “fake” connotation the President objects to. It makes no capability claim, which “super” cannot avoid. It has clean parallels: synthetic biology, synthetic data, both respectable fields named for how their products are made rather than how impressive they are. And it keeps the SI abbreviation. Accept the abbreviation, replace the semantics. Rhetorical judo.

Here is the governance point, and it is the reason this belongs on this blog. A term in a policy document should declare what it governs, not what it aspires to. Yesterday’s post argued that a boundary is only real when it is declared, checkable, and enforced. A label that changes while the definition underneath it does not move is the opposite of a declared boundary: it asserts without defining. The OSTP director’s 60-day homework is where the real fight happens, because the definition is the declaration. Until it lands, federal “SI” means exactly what “AI” meant, which means the rename governs nothing. It only frames.

One more thing worth noticing, because it is genuinely good news. The White House Accord on Super Intelligence, signed the same day by Trump and executives from Google, Anthropic, Meta, OpenAI, Nvidia, and xAI, is voluntary and only “morally binding” (with a typo, USA Today notes, in “Unites States”). But its substance is familiar to readers of my SOX-for-AI framework: companies commit to internal processes and controls, partnership with an independent external auditor for assessments, and a designated independent board committee for oversight. That is the assurance plane showing up in federal policy. Voluntary for now, but named correctly: controls, auditor, board. No category errors. If the vocabulary follows the accord rather than the rename, we get somewhere.

So call it what it is. Synthetic intelligence, built by people, governed by declared boundaries, checked by auditors. Names are governance. Choose them like it.