Decoder. plain-English AI glossary

ASI

▲ Risingay-ess-eye

Hypothetical AI that far surpasses human intelligence in all domains — a speculative concept that fuels both aspiration and existential risk concerns.

Think of it like

What a human brain is to an ant's brain, but more extreme: ASI to humans might be incomprehensibly vast.

Example

In science fiction, ASI rapidly solves climate change, invents new materials, redesigns biology — but might also decide humans are irrelevant.

How it actually works

ASI sits even further ahead of AGI: it's not just general intelligence, it's *superhuman* intelligence that compounds recursively. The idea: once you have AGI, it could improve its own algorithms, leading to an intelligence explosion ("singularity") where ASI emerges. This is speculative; we don't know if ASI is even possible (maybe there are hard limits to intelligence), and we definitely can't build it today. The framing shapes existential risk research: if ASI is achievable and misaligned, it's an extinction-level risk; if it's aligned, it could solve all human problems. The debate is intense and mostly speculative.

For product teams

A tail-risk scenario driving AI safety research; the timeline and probability are radically uncertain.

For engineers

A hypothetical AI system with vastly superhuman intelligence across all domains, capable of recursive self-improvement.

Related

Read anything AI without the jargon

Look up any term in plain English, or save terms as you read with the free Chrome extension.

Open DecoderAdd to Chrome