ONLINEAGENT_OPS 2026.Q3 HOME HUBSMODELSBUILDINGCOMPANIESFILMMAKINGSECURITYLIVING WITH ITSKILLSALL HUBS, ONE TERMINAL → ARTICLES CRAFT RECORD BLOG TOOLS SKILLS SEARCH /
HOME›AI MEANINGS›WHAT IS SUPERINTELLIGENCE (ASI)?
AI MEANINGS · MEANINGBEGINNER

What is superintelligence (ASI)? AI beyond human level

Superintelligence (ASI) means a hypothetical AI far smarter than the best humans at almost everything. Who claims it is coming, and what nobody knows.

READ5 min
WORDS1,070
SECTIONS8
TYPEMEANING
CHECKED5 OCT 26

Superintelligence, or ASI (artificial superintelligence), means a hypothetical AI that would be far smarter than the best humans at almost everything, from science to social skills. It is still hypothetical. Some AI companies say they are building it, others warn against it, and nobody knows whether or when it could happen.

◈ IN PLAIN TERMS

Think of the gap between a person and a cat. The cat is clever in its own way, but it cannot follow what we do in a laboratory or a parliament. Superintelligence is the idea of a machine that would stand that far above us. It is an idea about the future, not a product you can use today.

In a sentence: “The documentary asked whether superintelligence, an AI smarter than any person at almost everything, could ever be kept under control.”

What the word means

The plain point: beyond human, across the board, and so far only on paper.

IBM, the computer company, defines it as “a hypothetical software-based artificial intelligence (AI) system with an intellectual scope beyond human intelligence.”1 The philosopher Nick Bostrom gave a widely used definition in 1997: an intellect “much smarter than the best human brains in practically every field, including scientific creativity, general wisdom and social skills.”2

A 2023 Google DeepMind paper puts it at the top of its ladder of AI levels, as general AI that “outperforms 100% of humans”, and marks it “not yet achieved”.3

A simple example: the ladder

The plain point: superintelligence is the step after AGI, and both are beyond today’s narrow tools.

  • Narrow AI: good at one job, such as recognising faces. Exists today.
  • AGI: about as capable as people at most mental work. Debated, with no agreed test.
  • Superintelligence: far beyond the best people at nearly everything. Hypothetical.
TAKEAWAY

When someone says “superintelligence”, they are describing a future they expect or fear, not a system that has been built or tested.

Who claims what

The plain point: the loudest claims come from the companies building AI, and they are forecasts, not measurements.

OpenAI wrote in July 2023 that superintelligence “could arrive this decade”, and that it “could also be very dangerous”.4

Meta said in July 2025 that “Developing superintelligence is now in sight” and that it aims to bring “personal superintelligence to everyone.”5

Critics want it stopped for now. A public statement hosted by the Future of Life Institute calls for “a prohibition on the development of superintelligence” until there is scientific agreement it can be done safely, and public support.6

These are each organisation’s own statements. A company that sells AI has a stake in how powerful people think it will become.

What is unknown

The plain point: whether it is possible, when, and how anyone would keep it under control.

  • Whether it can be built. IBM notes that “Not all thinkers are aligned on the feasibility of something like an ASI”, partly because the brain is still not fully understood.7
  • When. Every date is a forecast. AGI meaning shows how far expert guesses spread for AGI, the earlier step.
  • How to control it. OpenAI wrote in 2023 that it had no solution for steering a superintelligent AI, and that today’s training methods rely on people being able to check the AI’s work.4 That research problem is part of AI alignment.

Where the word comes from

The word joins “super” and “intelligence”. The philosopher Nick Bostrom set out a definition in a paper first written in 1997: “an intellect that is much smarter than the best human brains in practically every field”.2

The paper was published in 1998, and Bostrom notes that his definition does not say what the superintelligence would be built from.8 The sources checked for this page do not say who used the word first.

Often confused with

  • AGI: AI roughly equal to people across most tasks. Superintelligence is a further claim: far beyond them. See AGI meaning.
  • A superhuman narrow tool: a program that beats every person at one game or one science task is not superintelligent, because it does nothing else. See narrow AI meaning.
  • Consciousness: being very capable and having feelings or awareness are separate questions. Does AI understand anything? covers the gap.

Questions people ask

Does superintelligence exist?

No. IBM calls it “hypothetical”, and the Google DeepMind levels paper lists artificial superintelligence as “not yet achieved”.13

What is the difference between AGI and ASI?

AGI means roughly human-level ability across most mental work; Stanford’s Institute for Human-Centered AI describes it as AI that “seeks broadly intelligent, context-aware machines”.9 ASI means beyond the best humans: in the DeepMind levels paper it is the top level of general AI, which “outperforms 100% of humans”.3 ASI is the bigger, further claim.

Learn more on this site

QUICK CHECK · 3 QUESTIONS

Pick an answer to see if you got it. Nothing is recorded.

1. Does superintelligence exist today?

  • No, it is still hypothetical
  • Yes, the biggest chatbots already are
  • Yes, but only inside research labs
Show the answer

Answer: No, it is still hypothetical. The page says IBM calls it hypothetical and a Google DeepMind paper lists it as not yet achieved.

2. Where does superintelligence sit on the page’s ladder of AI?

  • Below narrow AI, as an early first step
  • Level with AGI, just under a newer name
  • Above AGI, far beyond the best people
Show the answer

Answer: Above AGI, far beyond the best people. The page’s ladder goes narrow AI, then AGI, then superintelligence, which would be far beyond the best people at nearly everything.

3. How should you read AI companies’ claims about superintelligence?

  • As measurements from systems already tested
  • As forecasts from firms with a stake in them
  • As promises that are backed by law
Show the answer

Answer: As forecasts from firms with a stake in them. The page says the claims are forecasts, not measurements, and a company that sells AI has a stake in how powerful people think it will become.

Related terms: RLHF · Scaling laws · Singularity (technological singularity) · Supervised learning · Synthetic data · Training data

SOURCES
  1. 1IBM, What is artificial superintelligence?, published 14 Dec 2023, read at source 5 Oct 2026“Artificial superintelligence (ASI) is a hypothetical software-based artificial intelligence (AI) system with an intellectual scope beyond human intelligence.” ↑a“Artificial superintelligence (ASI) is a hypothetical software-based artificial intelligence (AI) system with an intellectual scope beyond human intelligence.” ↑b
  2. 2Nick Bostrom, How Long Before Superintelligence?, 1997, revised 1998, read at source 5 Oct 2026“By a ‘superintelligence’ we mean an intellect that is much smarter than the best human brains in practically every field, including scientific creativity, general wisdom and social skills.” ↑a“an intellect that is much smarter than the best human brains in practically every field” ↑b
  3. 3Morris et al. (Google DeepMind), Levels of AGI for Operationalizing Progress on the Path to AGI, arXiv:2311.02462, revised 24 Sep 2025, read at source 5 Oct 2026, Table 1“Level 5: Superhuman outperforms 100% of humans” … “Artificial Superintelligence (ASI) not yet achieved”. ↑a“Level 5: Superhuman outperforms 100% of humans” … “Artificial Superintelligence (ASI) not yet achieved”. ↑b“Level 5: Superhuman outperforms 100% of humans” … “Artificial Superintelligence (ASI) not yet achieved”. ↑c
  4. 4OpenAI, Introducing Superalignment, 5 Jul 2023, read at source 5 Oct 2026“But the vast power of superintelligence could also be very dangerous, and could lead to the disempowerment of humanity or even human extinction. While superintelligence seems far off now, we believe it could arrive this decade.” ↑a“Currently, we don’t have a solution for steering or controlling a potentially superintelligent AI, and preventing it from going rogue.” … “humans won’t be able to reliably supervise AI systems much smarter than us”. ↑b
  5. 5Meta, Personal Superintelligence, 30 Jul 2025, read at source 5 Oct 2026“Developing superintelligence is now in sight.” … “Meta’s vision is to bring personal superintelligence to everyone.” ↑
  6. 6Future of Life Institute, Statement on Superintelligence, undated page, read at source 5 Oct 2026 (the page says “Hosted by Future of Life Institute”): “We call for a prohibition on the development of superintelligence, not lifted before there is broad scientific consensus that it will be done safely and controllably, and strong public buy-in.”↑
  7. 7IBM, What is artificial superintelligence?, read at source 5 Oct 2026“Not all thinkers are aligned on the feasibility of something like an ASI.” … “the brain’s workings are still not fully understood making it difficult to recreate via software and hardware.” ↑
  8. 8Nick Bostrom, How Long Before Superintelligence?, International Journal of Futures Studies, vol. 2, 1998, revised 25 Oct 1998, read at source 12 Oct 2026the definition “leaves open how the superintelligence is implemented” ↑
  9. 9Stanford HAI, Artificial Intelligence Definitions, Sep 2020, read at source 5 Oct 2026 (as cited on narrow AI meaning): “Human-level AI, or Artificial General Intelligence (AGI), seeks broadly intelligent, context-aware machines.”↑

Morris et al. (Google DeepMind), Levels of AGI for Operationalizing Progress on the Path to AGI, arXiv:2311.02462, read at source 5 Oct 2026.

The comparison on this page (a person and a cat) and the three-step ladder are this site’s own ways of explaining, not claims from these sources.

Free printable AI cheat sheets, and AI news every week.

Sixteen one-page AI guides and checklists in one free PDF, no sign-up. And if you want to keep up, the newsletter covers what changed in AI each week, in plain English, with the sources.

SUBSCRIBE FREE ↗ GET THE CHEAT SHEET PACK
aipromptgeneer://$ about this siteAI PROMPT GENEERAI explained in plain English. Every claim links to its source, with the date it was checked.NEWSLETTERWhat changed in AI each week, in plain English.SUBSCRIBE FREE ↗AI PROMPT GENEER · MACHINE ASSISTANCE DISCLOSEDEVERY ARTICLE CARRIES ITS OWN CHECKED DATE