Artificial Superintelligence (ASI)

Artificial superintelligence (ASI), often just “superintelligence”, is a hypothetical intelligence far beyond any human’s. Philosopher Nick Bostrom’s widely used definition is “any intellect that greatly exceeds the cognitive performance of humans in virtually all domains of interest.”

Not the same as “Super Intelligence” (SI)

Since September 2026, the U.S. federal government has used “Super Intelligence” (SI) as its name for all AI, including spam filters and chatbots. That is a different idea. ASI describes a system that does not yet exist. Read SI vs superintelligence for the full comparison.

Kinds of superintelligence

Bostrom distinguishes three forms, which could combine:

• Speed superintelligence: thinks like a human mind, but much faster.
• Collective superintelligence: many minds working together that outperform any individual.
• Quality superintelligence: thinks better, not just faster, in ways humans can’t match.

How it might arrive

Most scenarios run through AGI first. A system that can do AI research could improve itself, an intelligence explosion, and reach superintelligence quickly. Others expect a slower climb limited by compute, data and energy. Forecasts are compared in When will superintelligence arrive?

Why it’s controversial

Supporters believe ASI could cure diseases, solve climate change and transform the economy. Critics worry a system smarter than its creators could be impossible to control. See existential risk and alignment. In 2025, a coalition including AI pioneers Geoffrey Hinton and Yoshua Bengio called for a prohibition on developing superintelligence until it can be done safely. In September 2026, Senator Bernie Sanders introduced a bill to ban it.