Approximating an artificial general intelligence or a general collective intelligence. Issue 3 (2nd May 2022)

• Record Type: Journal Article
• Authors: Williams, Andy E.
• Is Part Of: International journal of collaborative intelligence. Volume 2:Issue 3(2021)
• Page Start: 210
• Page End: 223
• Publication Date: 2022-05-02
• Subjects: general collective intelligence -- GCI -- collective intelligence -- artificial general intelligence -- AGI -- human-centric functional modelling; Computational intelligence -- Periodicals; Cognitive science -- Periodicals; Information science -- Periodicals; 006.305
• Languages: English
• ISSNs: 2051-7122

Abstract

Human-centric functional modelling has recently been used to define a model of artificial general intelligence hypothesised to have the capacity for human-like general problem-solving ability (intelligence), as well as a model of general collective intelligence (GCI) with the potential for exponentially greater general problem-solving ability than any individual in the group. Orchestrating the cooperation necessary to implement a functional model of AGI is hypothesised to require GCI to be reliably achievable. Furthermore, regardless of the collective benefits, investment in GCI by any large, centralised entity is generally misaligned with their individual incentives. This paper explores how a rudimentary AGI and a rudimentary GCI might be implemented through approximating the functions of each, in order to reduce the effort required to create systems that provide sufficient value to incentivise more sophisticated implementations to be developed over time in a decentralised, bottom up way that can be initiated by small teams.