Browse tools

The basics

Superintelligence

In short

Superintelligence is a hypothetical AI that would be far smarter than the best human minds in practically every field. It is a stated goal for some AI companies and a serious worry for others.

Also called
Artificial superintelligence, ASI, Superintelligent AI, Ultraintelligent machine
Topic
The basics
Level
Beginner
First described
I. J. Good described the "ultraintelligent machine" and the "intelligence explosion" in 19651. Nick Bostrom's 1998 paper "How Long Before Superintelligence?" gave the standard definition3.

The philosopher Nick Bostrom gave the definition most people use: "an intellect that is much smarter than the best human brains in practically every field"3. The definition says nothing about how it is built or whether it is conscious3.

The idea is older than the word. In 1965 I. J. Good wrote about an "ultraintelligent machine" that could "far surpass all the intellectual activities of any man"12. Such a machine could design even better machines, he argued, setting off an "intelligence explosion"12.

Bostrom separates two kinds. A weak superintelligence would think roughly as well as a person, only much faster. A strong one would be better in kind, not just in speed3.

It is not the same as a chatbot that beats people at some tests. The bar is being better than the best humans across practically every field3.

The term now appears in company plans and in campaigns against them. Meta's chief executive wrote in July 2025 that the company's vision is "to bring personal superintelligence to everyone"4. A statement hosted by the Future of Life Institute calls for a prohibition on developing superintelligence until there is broad scientific consensus that it can be done safely and controllably, and strong public buy-in5.

Example

Good's own warning sums up the stakes: the first ultraintelligent machine would be "the last invention that man need ever make", provided it is "docile enough to tell us how to keep it under control"2.

Why it matters

When AI leaders talk about building superintelligence, they mean systems that would outthink every person alive. Whether that can be done safely is one of the biggest open questions in AI policy.

Questions people ask

Is superintelligence the same as AGI?

Not quite. Superintelligence means being much smarter than the best human brains in practically every field3. Artificial general intelligence is a separate term, aimed at broad human-level ability rather than far beyond it.

Does superintelligence exist yet?

It is discussed as something in the future. Meta describes personal superintelligence as its vision4, and campaigners call for a ban on developing it until it can be done safely5.

Who came up with the idea of superintelligence?

I. J. Good wrote about an "ultraintelligent machine" and an "intelligence explosion" in 19651. Philosopher Nick Bostrom gave the modern definition in a 1998 paper3.

What is an intelligence explosion?

It is I. J. Good's idea that a machine cleverer than any person could design even better machines, which could then design better ones again, quickly leaving human intelligence far behind12.

Related terms

Artificial general intelligence Existential risk AI alignment AI safety P(doom) Frontier model Narrow AI

Sources

  1. Good, I. J. (1965), Speculations Concerning the First Ultraintelligent Machine, Advances in Computers vol. 6, pp. 31-88
  2. Quote Investigator, The First Ultraintelligent Machine Is the Last Invention That Humanity Need Ever Make (quotes Good 1965, p. 33)
  3. Bostrom, N. (1998), How Long Before Superintelligence?, International Journal of Futures Studies (author's revised version)
  4. Meta, Personal Superintelligence (letter from Mark Zuckerberg)
  5. Future of Life Institute, Statement on Superintelligence

Back to the glossary