Superintelligence
Also called ASI, Artificial Superintelligence, Machine Superintelligence
A hypothetical AI that would outperform the best humans at nearly every intellectual task, including science, strategy and creative work. No such system exists today.
What is superintelligence?
Superintelligence, often shortened to ASI (artificial superintelligence), describes an AI system that is far more capable than the best human minds across almost every intellectual task: scientific research, planning, programming, persuasion and creative work.
It is a hypothetical goal, not a product. No system available today meets it. The word is mostly used by researchers, AI labs and policy writers when they discuss where AI could go and what that would mean.
How does it relate to artificial intelligence?
"Artificial intelligence" covers every system that performs tasks that normally need human thinking, from spam filters to chatbots. Superintelligence is a much narrower idea: AI that goes well beyond human ability in general, not just at one task.
Some people now use the word as a name for the long-term direction of the field. The AI tools you can use today are still called AI.
How is it different from AGI?
- AI today: strong at specific tasks such as writing, coding or image generation, with real limits.
- Artificial general intelligence (AGI): a system that can learn and do most tasks a person can, at roughly human level.
- Superintelligence: a system that goes beyond human level across the board.
AGI is usually described as the step before superintelligence. Experts disagree about whether either will arrive, and when.
What is a simple example?
A strong chess program already beats every human player, but it can do nothing else. A superintelligent system would be like that chess program at every intellectual task at once: it could run the experiment, write the paper and design the next version of itself.
Why does it matter?
If a system far more capable than people were ever built, the goals it pursues would matter a great deal, which is why researchers study AI alignment and why some labs call for governance of such systems before they exist. For now it is a planning question, not a shopping one: to see what current AI agents and tools can do, browse the Ploba catalogue.