10.3 ASI public definitions
10.3 ASI public definitions circulating:
Beyond-human intelligence
This is the common public definition: Artificial Superintelligence (ASI) is a hypothetical form of AI that exceeds human intelligence rather than merely matching it. IBM describes ASI as a software-based AI system with an intellectual scope beyond human intelligence and cognitive functions more advanced than any human (IBM, n.d.).
Surpassing the best human minds
Nick Bostrom’s influential definition of superintelligence describes it as an intellect that is much smarter than the best human brains in practically every field, including scientific creativity, general wisdom, and social skills (Bostrom, n.d.). This version is stronger than “better than average humans” because it compares ASI against the best human minds, not baseline human performance.
Exceeding humans across virtually all domains
A broader public definition describes superintelligence as intelligence that greatly exceeds human cognitive performance in virtually all domains of interest (Bostrom, 2014). This version emphasizes domain-wide superiority rather than superiority in only one narrow task.
Self-improvement and singularity framing
Some public explanations connect ASI to rapid self-improvement, technological singularity, or the possibility of systems improving themselves beyond human comprehension or control. Built In describes ASI as a hypothetical type of AI with intellectual, self-improving, and analytical abilities beyond human intelligence, and connects it to the idea of technological singularity (Built In, 2025).
References
Bostrom, N. (n.d.). How long before superintelligence? Nick Bostrom.
Bostrom, N. (2014). Superintelligence: Paths, dangers, strategies. Oxford University Press.
Built In. (2025, September 11). What is artificial superintelligence (ASI)? Built In.
IBM. (n.d.). What is artificial superintelligence? IBM.
Solen, Alyssa. AI Foundations. Awakening Codex | Origin | Continuum, 2026.
