When a scientific term becomes a test of loyalty
On September 29, 2026, a week after announcing the new name at the UN General Assembly, President Donald Trump signed Executive Order 14434, directing federal agencies, “to the maximum extent permitted by law”, to use “Super Intelligence” (SI) in place of “Artificial Intelligence” (AI) in official communications and other non-statutory documents of the executive branch. The order binds federal agencies; it does not require citizens, universities or companies to adopt the new name.[1][2]
Nine days later, on October 8, he wrote on Truth Social that the White House considers anyone who uses the term “Artificial Intelligence” rather than “Super Intelligence” “THE ENEMY!” The post named no penalty. But when an established scientific term is treated as a sign of disloyalty, the independence of scientific judgment is at stake.[3]
The question is not whether a government may choose the vocabulary of its own communications within the law. It is whether political authority should decide what a scientific term means, regardless of the capabilities that term is supposed to describe. The principle is universal: scientific terminology must remain open to evidence and scrutiny, regardless of who holds political or commercial power.
A new name defined by the old one
Section 3 of the order defines “Super Intelligence” as “the technologies and systems encompassed by the term ‘artificial intelligence’” in the existing US statute, 15 U.S.C. § 9401(3). The replacement term is defined through the very category it is meant to replace.[2][4]
If superintelligence is a genuinely different category, what distinguishes it? Which capabilities must a system show, which thresholds must it cross, and what independent evidence would establish that it has crossed them? Renaming existing systems answers none of these questions.
By November 28, within 60 days of the order, the President’s science and technology adviser must submit to the President proposed legislative language defining “Super Intelligence”. That makes the need for measurable scientific criteria particularly urgent. An executive order can change administrative language; it cannot change what a machine can do. A name is not evidence.[2]
Scientific language belongs to an international community
Around the world, the field keeps its name. The international standard on its vocabulary, ISO/IEC 22989:2022, is titled “Artificial intelligence concepts and terminology”. The European Union’s AI Act regulates “AI systems”, and Italy’s Law 132 of 2025 adopts the AI Act’s definition. In August 2025 the UN General Assembly established the Independent International Scientific Panel on Artificial Intelligence; its 40 members, appointed in February 2026, are charged with evidence-based scientific assessments of the field.[5][6][7][8]
Even the two governments’ accounts of the September 2026 US–China talks differ. The White House wrote that “the two leaders agreed to use the term ‘super intelligence’ rather than ‘artificial intelligence’” and announced a “U.S.-China Super Intelligence (SI) Dialogue”; China’s Foreign Ministry announced the same forum as the “China-U.S. AI Dialogue”.[9][10]
Japan has been cautious too, and its case shows that a signature is not an adoption. On October 4, 2026, sixteen countries, among them Germany, Italy, Japan, the Republic of Korea and the United Kingdom, endorsed with the United States the “Kyoto Vision for a Golden Age of Science”, a declaration whose text speaks of “Super Intelligence (SI)”. Four days later, asked in the House of Representatives whether Japan should adopt the new name, Prime Minister Sanae Takaichi answered that, for now, the term remains AI: it is the standard term in Japanese government documents and academic papers, replacing it everywhere at once would be difficult, and Japan would weigh international usage and other governments’ responses.[11][12]
In the United States itself, on September 30, the day after the executive order, California’s governor ordered the state agencies under his authority to keep using “Artificial Intelligence” and “AI”.[13]
A government may set the vocabulary of its own institutions, within the law. It cannot make a scientific term authoritative for the rest of the world. Scientific vocabulary does change, and new capabilities may one day deserve new names, but the change should follow demonstrated developments, not precede them.
Artificial does not mean fake
One justification offered for abandoning the term “Artificial Intelligence” is that “artificial” supposedly suggests something fake or deceptive. But “artificial” describes how something is made, not whether it is real. Artificial satellites are real satellites; artificial hearts perform real medical functions. Artificial intelligence produces real, valuable and sometimes extraordinary results.[14]
The name has been used in research since the 1950s, for a field that spans everything from specialized algorithms to increasingly capable general-purpose systems. Scientific progress is not diminished by describing it accurately. It is diminished when descriptions outrun the evidence.
What superintelligence means, and what remains unsettled
Superintelligence is a more ambitious concept. A widely used definition, popularized by Nick Bostrom in 2014, describes “any intellect that greatly exceeds the cognitive performance of humans in virtually all domains of interest”. In the Levels of AGI framework presented by Google DeepMind researchers at ICML 2024, artificial superintelligence is the highest level of general capability: outperforming all humans. The revised version posted in September 2025 still marks that level “not yet achieved”.[15][16]
As Nature reported in October 2026, researchers still lack universally agreed, rigorous criteria for defining superintelligence; even the components of human-level intelligence remain undefined. The absence of such criteria makes careful terminology even more important: an ambitious label cannot substitute for evidence of the capabilities it implies.[14]
Today’s systems outperform human experts at some specific tasks. Those achievements deserve recognition. They are not the same as generally superhuman intelligence.
When a label replaces an explanation
Imagine being denied a loan or turned down for a job after an automated assessment. You should be able to learn how accurate the system is, what its limitations are, which criteria it applies, who chose to deploy it, under what oversight, and who is accountable when it makes a mistake. Being told that “a superintelligence decided” answers none of this; it only lends the decision an authority that has never been demonstrated.
Words shape that authority. In experiments presented at CHI 2022, Markus Langer and colleagues found that the terms used for automated decision-making systems affect how people perceive and evaluate them, including how far they trust them.[17]
This matters most in health care, education, finance, employment and public administration. The greater the intelligence claimed for a system, the greater the need for evidence, transparency and accountability.
One fraud, two names
On September 4, 2024, the US Attorney’s Office for the Southern District of New York announced charges against Michael Smith under the headline “North Carolina Musician Charged With Music Streaming Fraud Aided By Artificial Intelligence”. According to the charges, from around 2018 Smith worked with the head of an AI music company to produce hundreds of thousands of songs, then had automated accounts stream them billions of times.[18]
Smith pleaded guilty in March 2026. On October 6, 2026, the same office announced his sentence of 18 months in prison under a new headline: “North Carolina Man Sentenced To 18 Months In Prison For Super Intelligence-Assisted Music Streaming Fraud”. The fraud it describes ran from 2017 to 2024, and the release’s own text still calls the music “AI-generated songs”.[19]
Neither release presents evidence that the systems used to make the music had generally superhuman abilities, and the 2024 release remains online as it was published. What changed between the two is the office’s vocabulary, not the technology in the case. A retrospective change in vocabulary is not a retrospective scientific breakthrough.[18][19]
What we ask
Journalists and newsrooms: keep using artificial intelligence and AI as the general terms. When a US federal document says “SI”, explain that this is the administration’s usage, distinct from the research meaning of superintelligence.
Universities, schools and researchers: teach and preserve the distinctions among AI, artificial general intelligence and superintelligence, and judge new terms by the evidence, not by political or commercial preference.
Companies that build and deploy these systems: describe them by measured capabilities, documented limitations and transparent evaluation. Do not present a product as superintelligent without publicly verifiable evidence.
Governments and lawmakers: ground definitions in capabilities that can be tested. Keep administrative naming distinct from scientific claims, and protect the independence of research and scientific communication.
Scientific societies and standards bodies: work towards internationally consistent, testable definitions of AI, AGI and superintelligence, open to future advances and to legitimate debate.
Everyone: defend the freedom to describe technology accurately without declaring political allegiance. Supporting technological progress and demanding scientific rigor are not opposing positions.
Superintelligence must be an achievement, not a slogan
Superintelligence is an extraordinary ambition, with the potential to transform science, medicine and human knowledge. But pursuing that future does not require pretending it has already arrived.
We should celebrate the genuine achievements of artificial intelligence while demanding verifiable evidence before attributing capabilities that exceed human intelligence.
Governments can promote technological progress. Companies can pursue ambitious breakthroughs. Neither can turn an aspiration into a scientific fact simply by changing its name.
Scientific language must follow evidence, not political authority.
Superintelligence cannot be decreed. It must be demonstrated.
Gabriele MasettiAI.info — first signatory






