The short answer
“Super intelligence” now has two meanings. In US government language since 29 September 2026, Super Intelligence (SI) is simply the new name for artificial intelligence: it covers the same chatbots, image generators and models that federal law already calls AI. In AI research, superintelligence (often called artificial superintelligence, or ASI) means a hypothetical future system that is far smarter than humans in almost every field. No tool that exists today is superintelligent in the research sense.
Until this week, “super intelligence” was a term for philosophers, AI safety researchers and readers of Nick Bostrom. Then the US government adopted it as its new word for all AI, and millions of people started seeing “SI” in headlines. This guide untangles the two meanings and shows how SI relates to AI, AGI and ASI.
AI vs SI vs AGI vs ASI at a glance
| Term | What it means | Does it exist today? | Examples |
|---|---|---|---|
| AI (artificial intelligence) | Software that performs tasks that normally need human intelligence: writing, recognizing images, translating, predicting. | Yes | ChatGPT, Midjourney, Grammarly |
| SI (Super Intelligence, US federal term) | The US government’s new name for AI since 29 Sep 2026. Same legal definition as AI. | Yes, because it is the same thing as AI | Every AI tool, as named in federal documents |
| AGI (artificial general intelligence) | A system that can learn and do almost any intellectual task a person can, at a human level. | Debated; no agreed example | The stated goal of several AI labs |
| ASI (artificial superintelligence) | A system that greatly exceeds the best humans in nearly every field, including science, strategy and social skills. | No | Hypothetical |
Swipe the table sideways โ
Meaning 1: SI, the new US government name for AI
On 29 September 2026 President Trump signed the executive order “Inaugurating the Era of Super Intelligence”. It tells federal agencies to write “Super Intelligence” and “SI” wherever they used to write “Artificial Intelligence” and “AI” in new documents and websites. Crucially, the order says SI means the same technologies covered by the existing federal definition of artificial intelligence (15 U.S.C. 9401(3)). So under the order, a customer-service chatbot or a photo filter counts as SI.
The White House argues the word “artificial” undersells the technology. Critics argue the new name oversells it, because it borrows a term researchers reserve for systems far beyond today’s. Either way, the legal meaning has not changed. A new federal definition is due to be proposed by 28 November 2026. Read the full breakdown in AI is now SI: the order explained.
Meaning 2: superintelligence in AI research
In research, superintelligence describes an intellect that is much smarter than the best human brains in practically every field. The idea goes back to the mathematician I. J. Good, who wrote in 1965 about an “ultraintelligent machine” that could design even better machines, setting off an “intelligence explosion”. Philosopher Nick Bostrom made the term mainstream with his 2014 book Superintelligence: Paths, Dangers, Strategies, which argued that such a system could be hard to control and that its goals must be aligned with ours before it is built.
Researchers usually describe three possible forms:
- Speed superintelligence: thinks like a human, but thousands of times faster.
- Collective superintelligence: many systems working together, smarter as a group than any human organization.
- Quality superintelligence: thinks in ways that are simply better than human thinking, not just faster.
Big AI companies use this meaning when they talk about their long-term goals. Meta renamed its AI division “Meta Superintelligence Labs” in June 2025, and Microsoft announced a “humanist superintelligence” effort in November 2025. These are ambitions, not products you can use.
Narrow AI, general AI and super AI
A common way to explain the levels is a three-step ladder:
- Narrow AI (weak AI): very good at one kind of task. Almost every tool in our directory is narrow AI: a transcription tool transcribes, an image generator makes images.
- General AI (AGI, strong AI): as capable as a person across most tasks, able to learn new ones without being rebuilt.
- Super AI (ASI): beyond the best humans at almost everything.
Today’s large language models such as Claude, Gemini and ChatGPT sit somewhere between the first and second steps: broad, but still unreliable at many things people do easily. Calling them “SI” in government documents does not move them up the ladder.
▶ Watch the Claude video review (0:37)
▶ Watch the ChatGPT video review (0:31)
Is there any real super intelligence example today?
Not in the research sense. Some systems are superhuman at one narrow job, for example chess and Go engines, protein-structure prediction, or reading thousands of documents in seconds. That is narrow superhuman performance, not general superintelligence. If someone sells you an “SI tool” today, it is an AI tool under a new label, which is fine, as long as you judge it by what it actually does.
How to read “SI” when you see it
- On a .gov website, in a federal tender or in White House material: SI = AI. Same law, same meaning.
- In an AI safety paper or a lab’s mission statement: superintelligence = ASI, a hypothetical future system.
- On a product page: almost certainly marketing for an ordinary AI tool. Check our SI tools list for tested options.
- Elsewhere: SI can also mean the International System of Units, Sports Illustrated or Slovenia’s .si web domain.
Frequently asked questions
What is super intelligence?
It has two meanings. In US federal language since 29 September 2026, Super Intelligence (SI) is the official new name for artificial intelligence. In AI research, superintelligence means a hypothetical system far smarter than humans in nearly every field.
What is the difference between AI and SI?
Legally, none. The executive order defines SI using the existing federal definition of AI. SI is the term federal agencies must now use; AI is the term everyone else still uses.
Is SI the same as ASI?
No. SI in the executive order covers all of today’s AI. ASI, artificial superintelligence, is a research term for a future system that would surpass humans at almost everything. No ASI exists.
What is artificial superintelligence (ASI)?
ASI is a hypothetical AI that exceeds the best human performance in practically every area, including science, creativity, planning and social skills. It is the level above artificial general intelligence (AGI).
Who wrote Superintelligence: Paths, Dangers, Strategies?
The philosopher Nick Bostrom, published in 2014. It is the book that brought the idea of superintelligence and its risks to a wide audience.
Does super intelligence exist yet?
Only as a new name. Under the US order, today’s AI is called SI. A superintelligence in the research sense, smarter than humans across the board, does not exist yet.
Paid remote work
Open remote AI-training roles
AI companies pay experts by the hour to test and improve AI. Remote and flexible. These roles are open now.
1,538 open roles ยท updated today
Get new roles like these by email
One short email each morning with new roles open in your country. Free. Unsubscribe any time.
Where to go next
Three routes, picked for what you just read.
See Zekai first in Google
The weekly AI briefing for your profession
One weekly email: the AI changes that actually affect your profession โ tools, deals, and what to do about them.

