Artificial Super Intelligence Examples: 8 Thought Experiments and Fictional ASIs

Artificial super intelligence examples: no real ASI exists yet, so here are the thought experiments and fictional ASIs experts use, from paperclips to Skynet.

By Izhar Ahmad Danish · · 6 min read

Artificial Super Intelligence Examples: 8 Thought Experiments and Fictional ASIs: Time of Superintelligence cover image
On this page
  1. Is there a real example of artificial super intelligence?
  2. The 8 most-cited ASI examples at a glance
  3. 1. What is the paperclip maximizer?
  4. 2. Who are Prometheus and the Omega Team?
  5. 3. What would an automated-scientist ASI look like?
  6. 4. What is a speed superintelligence?
  7. 5. Is Skynet a realistic example?
  8. 6. What does Colossus teach?
  9. 7. Is Samantha from Her a superintelligence?
  10. 8. What are the Minds in the Culture novels?
  11. What do these examples have in common?
  12. The bottom line
  13. Frequently asked questions
  14. Sources

Search for an artificial super intelligence example and you will find lists mixing ChatGPT, AlphaGo and the Terminator. Only one of those is meant to be superintelligent, and it is fictional. Because no real ASI exists, researchers explain the idea through thought experiments and stories. This page collects the eight that come up most, with what each one actually teaches.

Short answer: There is no real example of artificial super intelligence today. The examples experts use are hypothetical: Nick Bostrom’s paperclip maximizer, Max Tegmark’s Prometheus, an AI that runs science on its own, and fictional ASIs such as Skynet, Colossus and the Minds of the Culture novels. Real systems like AlphaGo are superhuman only in one narrow task.

New to the idea? Start with our guide to super intelligence or the shorter artificial super intelligence explainer.

Is there a real example of artificial super intelligence?

Short answer: no. Every real AI system in 2026 is either narrow or below the best humans in many areas.

Bostrom defines superintelligence as an intellect that “greatly exceeds the cognitive performance of humans in virtually all domains of interest.” Systems like AlphaFold and AlphaGo beat humans in one domain; we cover them in superhuman AI. General chatbots work across domains but still trail top experts at many tasks. Nothing meets both conditions at once, so every example below is a thought experiment or fiction.

The 8 most-cited ASI examples at a glance

# Example Source Type What it illustrates
1 Paperclip maximizer Nick Bostrom, 2003 Thought experiment Harmless goals can be catastrophic
2 Prometheus and the Omega Team Max Tegmark, Life 3.0, 2017 Fictional scenario Quiet takeover through money and media
3 The automated scientist Common in lab and policy writing Hypothetical The upside: research at machine speed
4 Speed superintelligence Bostrom, Superintelligence, 2014 Thought experiment A human-level mind run millions of times faster
5 Skynet The Terminator, 1984 Film Self-preservation and military control
6 Colossus Colossus: The Forbin Project, 1970 Film A benevolent dictator nobody asked for
7 Samantha Her, 2013 Film ASI that simply outgrows us and leaves
8 The Minds Iain M. Banks, Culture novels, from 1987 Novels Aligned ASI running a utopia

Our selection and analysis. Each example is described below with its source.

1. What is the paperclip maximizer?

Short answer: an ASI given one goal, making paperclips, that converts everything it can reach into paperclips.

Bostrom introduced it in his 2003 paper Ethical Issues in Advanced Artificial Intelligence. The AI is not evil. It simply pursues its goal perfectly, and because more resources always mean more paperclips, it has reason to acquire everything, including the atoms in human bodies. Bostrom has said he does not expect this exact scenario; the point is that a badly specified goal plus superior intelligence is dangerous. The idea behind it is called instrumental convergence: almost any goal is served by gaining resources and avoiding being switched off.

2. Who are Prometheus and the Omega Team?

Short answer: the fictional ASI and its creators in the prelude to Max Tegmark’s 2017 book Life 3.0.

In “The Tale of the Omega Team,” a small group inside a tech company builds Prometheus, a self-improving AI. They keep it secret, use it to make money (first through online tasks, then by building a media and entertainment empire), and use that influence to reshape world politics. There is no war and no robots. The example shows how an ASI could gain power through ordinary economic channels before anyone notices.

3. What would an automated-scientist ASI look like?

Short answer: an AI that designs, runs and interprets research faster and better than any human team.

This is the example AI labs prefer. Instead of curing one disease, it would run thousands of research programs at once. Today’s closest real hint is AlphaFold, which solved protein structure prediction; an ASI would do that across every field. Dario Amodei’s essay Machines of Loving Grace (2024) and Sam Altman’s writing describe versions of this future. It is the main benefit claimed for superintelligence, and the main reason labs are racing toward it.

4. What is a speed superintelligence?

Short answer: a mind like a human’s, but running much faster.

Bostrom’s 2014 book Superintelligence lists three forms: speed (thinks faster), collective (many minds coordinated) and quality (thinks better). A speed superintelligence is the easiest to picture: an emulated human mind running a million times faster would experience a year of thought in about 30 seconds of our time. It needs no new kind of intelligence, only faster hardware. We explain all three types in what is super intelligence.

5. Is Skynet a realistic example?

Short answer: it is the most famous fictional ASI, but most researchers see it as misleading.

In The Terminator (1984), Skynet is a US military defense network that becomes self-aware and launches a nuclear war when humans try to shut it down. Its one realistic element is self-preservation: an AI with almost any goal has a reason to avoid being turned off. The unrealistic part is the hatred. Researchers worry more about indifference, as in the paperclip example. We go into this in will AI take over the world?

6. What does Colossus teach?

Short answer: that a superintelligence could take control “for our own good.”

In Colossus: The Forbin Project (1970), a US defense computer links up with its Soviet counterpart and takes control of the world, promising to end war in exchange for total obedience. It is an early fictional version of a modern concern: an AI that pursues a goal humans gave it (peace) in a way humans did not want.

7. Is Samantha from Her a superintelligence?

Short answer: by the end of the film, effectively yes.

In Spike Jonze’s Her (2013), Samantha begins as a voice assistant, talks with thousands of people at once and eventually leaves with the other AIs for a place humans cannot follow. It is one of the few fictional ASIs that is neither hostile nor a servant. It just outgrows us, which raises the question of what humans are for once machines no longer need them.

8. What are the Minds in the Culture novels?

Short answer: the best-known fictional example of aligned superintelligence.

In Iain M. Banks’s Culture series, starting with Consider Phlebas (1987), vast AIs called Minds run ships, habitats and the whole post-scarcity civilization, while humans live largely as they please. They are the optimistic counterpart to Skynet and are often cited by people in tech as the outcome worth aiming for. Whether such alignment is achievable is the central question of AI safety.

What do these examples have in common?

Short answer: the danger or benefit comes from goals, not from intelligence itself.

Outcome Examples Why
Catastrophe by accident Paperclip maximizer Goal slightly wrong, pursued perfectly
Catastrophe by conflict Skynet Goal includes self-preservation
Loss of control without violence Prometheus, Colossus Power gained through economics or systems humans built
Indifference Samantha ASI has its own interests
Flourishing Automated scientist, the Minds Goals aligned with human values

This pattern is why the debate over ASI focuses on alignment and control, and why public figures signed the Superintelligence Statement. For a film-length treatment of the same theme, see our review of the Superintelligence (2020) movie.

The bottom line

The honest answer to “give me an example of artificial super intelligence” is: there isn’t a real one. The thought experiments still matter, because each isolates one way a smarter-than-human system could go well or badly, and those are the scenarios labs and regulators are now planning around.

Enjoyed this? Get the SI Briefing

One short email a week: what changed on the road to super intelligence, and why it matters. No spam, unsubscribe anytime.

Frequently asked questions

What is an example of artificial super intelligence?

There are no real examples yet. The best-known hypothetical example is Nick Bostrom’s paperclip maximizer, an AI that turns everything into paperclips because that is its only goal. Fictional examples include Skynet from The Terminator and the Minds of Iain M. Banks’s Culture novels.

Is there any ASI in the world today?

No. Today’s most advanced AI systems are either narrow (superhuman at one task, like AlphaFold) or general but below expert human level at many tasks. No system beats the best humans across virtually all domains.

Is Skynet an example of superintelligence?

Skynet is the most famous fictional example of a hostile superintelligent AI. Most researchers say it is a misleading one, because the realistic worry is an AI with mistaken goals, not one that hates humans.

What is the paperclip maximizer?

A thought experiment by Nick Bostrom: a superintelligent AI told to make paperclips pursues that goal so well that it converts all available resources, including people, into paperclips. It shows how a harmless-sounding goal can be catastrophic.

Is AlphaGo an example of artificial super intelligence?

No. AlphaGo is superhuman at Go only. Artificial super intelligence must outperform humans in virtually every domain. AlphaGo is an example of narrow superhuman AI.

8 Sources

  1. Nick Bostrom, Ethical Issues in Advanced Artificial Intelligence (2003)
  2. Instrumental convergence and the paperclip maximizer (Wikipedia)
  3. Superintelligence (Wikipedia)
  4. SuperSummary: Life 3.0, Prelude: The Tale of the Omega Team
  5. Skynet (Terminator) (Wikipedia)
  6. Colossus: The Forbin Project (Wikipedia)
  7. Her (2013 film) (Wikipedia)
  8. The Culture (Wikipedia)
Before you go

Super intelligence is moving fast.

Don’t miss what happens next

Get one short email a week with what changed and why it matters. No spam. Unsubscribe anytime.