Redefining Artificial Superintelligence: From Mystified Genius to Observable Conditions

Existing definitions of artificial superintelligence (ASI) share a common structural defect: they define ASI relativeto human cognitive baselines that are themselves unmeasurable. This paper identifies the root cause—thesystematic mystification of historical genius, which renders any definition built on ‘surpassing the best humanminds’ structurally incoherent—and proposes an alternative. We critically examine the four dominant definitionsin the field (Bostrom 2014; Legg & Hutter 2007; Chollet 2019; Goertzel 2007) and show that each fails to providean observable, testable criterion for superintelligence. We then propose a four-condition definition groundedentirely in observable AI capabilities: processing speed, collective knowledge integration, scientific creativity, andgeneral wisdom under complexity. Two of these conditions are already satisfied by current AI systems; theremaining two define the actual frontier of ASI research. This reframing shifts the definitional baseline from thehuman side to the AI side, yielding a definition that is measurable, progress-trackable, and independent ofunmeasurable claims about human cognitive limits.