Trump Says U.S. Will Call AI 'Super Intelligence' and Why Wording Matters
President Donald Trump told the United Nations General Assembly on Tuesday morning that the United States is officially renaming artificial intelligence to "super intelligence." The Verge
Trump said all United States documents, and hopefully documents around the world, will change from the word "artificial" to "super" for super intelligence. He said "artificial" makes intelligence sound fake and called "super intelligence" more accurate and better sounding.
The remarks sit alongside other recent statements on AI governance. Trump has said the United States already has guardrails in place to regulate and prosecute AI companies. Reuters He has also said he plans to appoint a new artificial intelligence adviser known as an "AI czar." Reuters
Looking at what this means for people who build, buy and regulate these systems, language works like shared infrastructure. Model cards, which are short disclosure documents that describe what a model does and how it was tested, plus purchase requirements, compliance filings, export controls and internal risk reviews all rely on common terms. Change a core term and every reference below it needs reinterpretation. That creates extra work and uncertainty about scope.
In my view the proposed swap adds friction because it mixes two different ideas. Artificial intelligence is the name for a field of methods and systems, or how machines are built to perform tasks. Super intelligence is normally used for a capability level, or how powerful those systems are. Safety tests, access controls and deployment rules depend on keeping that distinction.
Worth flagging for documentation owners is the problem of versioning. Federal forms, guidance, contract clauses and technical standards point to each other and do not update by declaration alone. If one set of papers changes wording while others do not, vendors, auditors and lawyers must work with two vocabularies at once until the language lines up. That is manageable, but it is not costless.
The broader context here is adoption. AI tools are now ordinary enterprise software, inside help desks, code editors, call centers, design tools and data analysis workflows. I watched my own kids treat autocomplete and voice assistants as background utilities years ago. Business users have now done the same with larger models. We have seen this pattern before, when the PC and the commercial internet moved from novelty to utility and arguments shifted from whether the technology worked to what to call it and who defines it.
In my view, over the longer arc, clearer public language would still help. Artificial never meant fake in engineering use. It meant built, as opposed to biological. Super suggests a performance threshold that most systems in use today do not claim to reach. The more useful path is sharper use of existing terms for narrow systems, general-purpose frontier models, and any hypothetical successors, rather than one replacement label. That kind of precision helps buyers write requirements, helps engineers set test boundaries, and helps regulators write rules that can be enforced.
