On September 19, 2026, President Trump posted a poll on X asking which name should replace "Artificial Intelligence": Superior Intelligence, Extreme Intelligence, or Supreme Intelligence — two of which abbreviate right back to "SI," and none of which change a single line of code, a single training run, or a single regulation. The poll pulled in more than 140,000 votes within a day and instantly became the most-discussed AI story on the platform, ahead of genuinely consequential news the same week: Newsom's AI kill-switch executive order, a Senate floor fight over the Frontier AI Act, and Trump's own rejection of an industry-wide AI pause.
This is not the first time someone with power over AI's public image has tried to rename it. It is at minimum the fourth. Every prior attempt — commercial, academic, and now political — ran into the same wall: the name was never the problem. Here is the actual history, and why "artificial intelligence" has now outlasted 70 years of people trying to replace it.
TL;DR
| Question | Answer |
|---|---|
| What did Trump propose? | Renaming "Artificial Intelligence" to Superior Intelligence, Extreme Intelligence, or Supreme Intelligence, via an X/Truth Social poll |
| Who actually coined "artificial intelligence"? | John McCarthy, in his 1955 proposal for the 1956 Dartmouth Summer Research Project |
| Has AI been renamed before? | Yes — repeatedly, during both AI winters, as "machine learning," "cognitive computing," "informatics," and "knowledge engineering" |
| Did any rename actually replace "AI"? | No. Every alternate label ran in parallel and "artificial intelligence" survived as the umbrella term |
| What's the real objection to the poll's timing? | It landed one day after Newsom's kill-switch order and hours before Trump's own "AI Force" and AI czar announcement — critics read it as a branding exercise displacing a policy one |
| Does the new name change anything technical? | No — models, training methods, and capabilities are unaffected by what the field is called |
What Trump actually posted
Trump's post, shared across X and Truth Social, argued that "Artificial Intelligence" is "inaccurate, and very ineloquent, relative to AI," and proposed three replacements:
"A far more elegant and accurate description of this new phenomena would be Superior Intelligence (SI) or, Extreme Intelligence (EI) or, Supreme Intelligence (SI)... VOTE!"
The poll drew over 145,000 votes with a day still left on the clock, and quote-tweets ranged from earnest ("Rename it 'American Intelligence' and we can keep it as AI") to religious objection over calling a machine "Supreme," to the flatly dismissive: one widely shared reply read, "$6.50 for diesel and you want us to vote on what we should call computers?" Elon Musk weighed in with a one-line aside — "It will be incomprehensibly smart to us" — reframing the thread away from branding toward capability, which is closer to the actual stakes than any of the three options on the ballot.
None of the three proposed names describe anything different about how these systems work. A transformer trained with reinforcement learning from human feedback does not become more "supreme" if you change its category label — a point explainx.ai's guide to the field's real history makes clear: the term has always described a broad, shifting basket of techniques, not a fixed level of capability.
Who actually coined "artificial intelligence" — and why
The term dates to 1955, when Dartmouth mathematics professor John McCarthy drafted the proposal for what became the 1956 Dartmouth Summer Research Project on Artificial Intelligence — the workshop historians now call "the Constitutional Convention of AI." McCarthy organized it with Marvin Minsky (Harvard), Nathaniel Rochester (IBM), and Claude Shannon (Bell Labs), and it was McCarthy specifically who chose the words "artificial intelligence" to describe the field they were founding.
The workshop itself was a modest technical event — it did not produce the "thinking machine" McCarthy had envisioned, and few concrete results came out of that summer. But the name outlived the meeting by 70 years and counting, through two collapses in AI funding and hype, multiple architectural revolutions, and now a presidential poll trying to replace it.
Notably, even McCarthy came to have doubts about his own word choice. Sheffield AI professor Noel Sharkey has said McCarthy told him he wished he had picked "computational intelligence" instead — a term with less baggage about matching or exceeding human general intelligence, which was never what most AI research actually claimed to deliver.
The AI winters already tried this — twice
Trump's poll is being treated online as a novelty, but the field has been renamed before, under much more serious pressure than an X poll. During the first AI winter (roughly 1974–1980), triggered by the UK's 1973 Lighthill Report cutting government AI funding, and again during the second AI winter (roughly 1987–1993), when commercial expert systems collapsed under their own maintenance costs, "artificial intelligence" became a genuinely toxic label — not an ineloquent one, a financially dangerous one.
Researchers and funding applicants responded exactly the way you'd expect a stigmatized brand to be handled: they rebranded.
| Era | What people called the same work instead of "AI" | Why |
|---|---|---|
| Post-1974 (first AI winter) | "Informatics," "pattern recognition," "computational intelligence" | Grant reviewers penalized proposals using the tainted "AI" label |
| Post-1987 (second AI winter) | "Knowledge engineering," "decision support," "expert systems" (as a neutral-sounding subcategory) | Investors associated "AI" with the failed promises of the 1980s expert-systems boom |
| Both winters, ongoing | "Machine learning" | Framed as a narrower, empirically grounded discipline distinct from "AI's" grander claims |
| 2011–2022, commercially | "Cognitive computing" (IBM, marketing Watson) | IBM wanted distance from AI hype cycles while still selling an AI-powered product |
The irony historians point to: the most productive research period for what we now call deep learning happened while the word "AI" was out of fashion. Backpropagation, the technique that made multi-layer neural networks trainable, was refined and popularized in 1986 — squarely inside the second AI winter, under the "machine learning" and "neural networks" banners, specifically because "AI" carried too much baggage to attach to a funding request.
None of those rebrands displaced "artificial intelligence" once the field recovered. "Machine learning" survived as a genuinely useful narrower term (it describes a method, not the whole field). "Cognitive computing" quietly died with IBM Watson Health, sold off in 2022 for a fraction of its promised value, replaced by the watsonx platform. "Informatics" and "knowledge engineering" became historical footnotes. "Artificial intelligence" outlasted every euphemism invented to avoid saying it.
Why the name survives: it was never the name's fault
Both AI winters were caused by overpromising relative to delivered capability, not by the phrase "artificial intelligence" itself. Lighthill's 1973 report didn't fault AI researchers for a bad job title — it faulted them for failing to scale symbolic reasoning systems beyond toy problems. Expert systems collapsed under real maintenance costs and brittleness, not unfortunate branding.
That distinction matters for reading Trump's 2026 poll correctly. "Superior," "Extreme," and "Supreme" are marketing adjectives layered onto the same underlying acronym problem the field has faced since 1956: the public keeps expecting AI to mean general, human-equivalent (or superior) reasoning, while most deployed systems remain narrow statistical pattern-matchers wrapped in agent scaffolding. Renaming the category "Superior Intelligence" doesn't close that gap — if anything, it widens the expectations gap the way "cognitive computing" widened it for IBM Watson in oncology, a promise the tech ultimately could not keep at scale.
For anyone building products in this space, the actual lesson from two AI winters is not about vocabulary. It's what always causes the hype cycle to snap back: claims about system capability that outrun independent verification. That's precisely the fight playing out in 2026's pace-the-frontier debate and Frontier AI Act negotiations — a fight a naming poll cannot settle either way.
The timing: a poll next to a kill-switch order and an "AI Force"
The renaming poll did not land in a vacuum. It arrived exactly one day after California Governor Gavin Newsom signed Executive Order N-9-26 on September 18, 2026, directing the state's Government Operations Agency to complete a 60-day study on a mandatory "kill switch" for frontier AI models that "go rogue," alongside independent auditors embedded at large labs. Newsom's office framed the order bluntly: "The federal government's abject failure to create any form of meaningful AI oversight or accountability should alarm every American, especially when AI CEOs themselves are begging for regulation" — accusing Washington of being "asleep at the wheel" on AI safety while California moved unilaterally, building on the state's earlier SB 813 and AB 1405 AI audit laws.
Hours after the renaming poll, Trump posted a second, separate announcement: he would form an "AI Force," explicitly modeled on the Space Force he created during his first term, and would soon name an "AI czar" — adding that "only High I.Q. individuals need apply." He framed AI as potentially "25% of our Country's GDP" and, in the same breath, rejected new regulation: "criminal and civil laws are sufficient to address misconduct by AI companies." That statement follows Trump's earlier characterization of industry safety warnings — including Dario Amodei's "Pace the Frontier" essay — as a "hoax," made both in a post and live on a call to the All-In Summit.
Lined up chronologically, the week reads as three competing signals rather than one coherent policy:
- Sept 18 — Newsom orders a study on mandatory kill switches and onsite auditors, calling federal inaction a national alarm.
- Sept 19 (morning) — Trump posts the renaming poll: Superior, Extreme, or Supreme Intelligence.
- Sept 19 (later) — Trump announces an "AI Force" and forthcoming AI czar, while reiterating that no new regulation is needed.
Critics — echoing the same "distraction" framing California officials have used all year, including in the kill-switch order's own press language — argue that a naming poll occupying the news cycle the same week as a real regulatory fight is exactly the kind of showmanship that displaces substance rather than adding to it. Renaming AI requires no legislation, no agency rulemaking, and no vote in Congress; a kill-switch mandate requires all three, which is precisely why one produced a viral poll and the other produced a 60-day study.
What people are asking about the poll
Does this actually rename AI, or is it symbolic?
Purely symbolic. An X poll has no legal or regulatory force. Even if "Superior Intelligence" or "Supreme Intelligence" won every vote, "artificial intelligence" would remain the term used in statutes (like the EU AI Act), academic literature, and product documentation until an entirely separate, much slower process of institutional adoption happened — the same process that let "machine learning" survive as a term of art while "cognitive computing" did not.
Why do "Superior Intelligence" and "Supreme Intelligence" both abbreviate to "SI"?
That appears to be either an oversight or a deliberate wink — Trump's post did not address the overlap. Commentators noted the redundancy immediately; one popular reply suggested dropping "Superior" for "Super Intelligence" to avoid the collision, which ironically collides with the pre-existing, decades-old technical term "superintelligence" already used in AI safety literature (see Bengio and Bannon's superintelligence-ban coalition and Bernie Sanders' Artificial Superintelligence Act) to describe a specific, more advanced hypothetical capability tier — not a synonym for AI generally.
Has a government ever successfully renamed a scientific field?
Rarely, and mostly at the margins. "Global warming" partially gave way to "climate change" in some official usage, but both terms remain in active parallel use decades later — the same pattern AI's own history shows. Terminology shifts driven by politics or PR tend to add a competing term rather than erase the original; nobody stopped saying "artificial intelligence" because IBM said "cognitive computing," and nobody will stop because of a poll.
What's the actual stakes if AI branding shifts anyway?
Low, for the technology; not low, for public perception and regulation. If "Superior Intelligence" or similar language entered official use, it could subtly reframe policy debates — a name implying inherent superiority sets different public expectations than a name implying mere artificiality, potentially affecting how urgently voters expect Congress to regulate it. That's a genuine, if second-order, effect; it's just not the effect Trump's post argues for.
Where can I read the primary sources?
Start with Trump's original poll coverage, the AI Force and AI czar announcement, and the Dartmouth proposal document itself, still hosted by Stanford. explainx.ai's full 1950–2026 AI history timeline covers the Lighthill Report, both AI winters, and the deep learning revolution in more depth than fits here.
Steel-manning both sides
The case for taking the poll seriously as more than a joke: Names do shape public perception, and "artificial" genuinely does undersell what 2026-era frontier models do compared to 1956's ambitions. McCarthy's own regret about "computational intelligence" shows even the term's inventor thought a more precise label existed. A poll, however unserious in form, is at least a low-cost way of surfacing that the current name annoys people across the political spectrum for different reasons — accuracy on one side, marketing hype on the other.
The case against: Every one of the three proposed names (Superior, Extreme, Supreme) leans toward overstatement rather than precision — the opposite direction from McCarthy's preferred fix. None of them would change what a single AI system actually does, and the poll ran the same week as a real regulatory fight over kill switches and frontier-model oversight that a name change cannot substitute for. History's verdict on 70 years of AI rebrand attempts is unambiguous: the label was never the problem, and no successful rename has ever displaced "artificial intelligence" — it has only added a parallel term until the field's actual capability caught up with, or fell short of, the original name.
Related reading
- The History of Artificial Intelligence: From Turing's 1950 Test to AGI in 2026
- Newsom's AI Kill Switch Executive Order N-9-26
- California's SB 813 and AB 1405 AI Audit Laws
- Trump and Speaker Johnson Reject an AI Pause
- Pace the Frontier: Baker, Burry, Trump, and Harris React
- The Frontier AI Act Floor Vote
- AI Regulation: EU AI Act and US Policy Complete Guide
- Bernie Sanders' Artificial Superintelligence Act
For the primary sources behind this post: Dartmouth's own account of coining "AI", the original 1955 Dartmouth proposal, and The Hill's coverage of the poll.
This post reflects reporting and public statements available as of September 20, 2026. Poll results, vote totals, and any follow-up naming announcements may change after publication — check the linked primary sources for the latest figures.