By Roman Aglaganov
On 27 October President Trump posted that “in short order” his government would be “officially BANNING the term, ‘AI/Artificial Intelligence,’ in all relevant documents, and everything else, due to its inaccuracy.” He wants “Super Intelligence”, or SI, instead. An executive order he signed on 29 September already tells the executive branch to use those terms, arguing that they “more appropriately” capture “the promise, potential, and rapidly advancing capabilities of these technologies.”
A name that oversells
The president is right that “artificial intelligence” is an imperfect name. It was coined more than 70 years ago, and it has always promised more than it describes. But “Super Intelligence” makes the problem worse. In the research world, superintelligence means systems far smarter than people at almost everything. Today’s models are impressive and uneven: they can draft a contract and then miscount the letters in a word.
Words shape decisions. A procurement officer told that a tool is “super intelligent” may trust it more than the evidence allows. Many of this week’s stories are about exactly that gap: agents that need checking, AI-made proofs awaiting expert review, chatbots that must be stopped from pretending to be human. Calling these systems super intelligent invites people to skip the checking.
Precision matters in law
Government documents are not slogans. Laws, contracts and safety rules depend on definitions, and “AI” is the term used in existing statutes, in the EU’s AI Act, and in standards. Swapping it for a vaguer, grander term in some federal documents, but not in law or in the rest of the world, risks confusion over what a rule actually covers.
There is also a limit to what a president can rename. The order applies to the executive branch “to the maximum extent permitted by law”. It cannot make journalists, scientists or companies change their words, and most AI companies still call their products AI.
Spend the effort elsewhere
If the goal is to show that America takes the technology seriously, there are better ways: fund the government’s testing of AI models, set clear rules for what agents may do on people’s behalf, and publish what federal agencies learn when they deploy these tools. Those things would make AI safer and more useful. A new name will not change what the systems can do.
