General intelligence disentangled via a generality metric for natural and artificial intelligence

Abstract Success in all sorts of situations is the most classical interpretation of general intelligence. Under limited resources, however, the capability of an agent must necessarily be limited too, and generality needs to be understood as comprehensive performance up to a level of difficulty. The...

Descripción completa

Guardado en:
Detalles Bibliográficos
Autores principales: José Hernández-Orallo, Bao Sheng Loe, Lucy Cheke, Fernando Martínez-Plumed, Seán Ó hÉigeartaigh
Formato: article
Lenguaje:EN
Publicado: Nature Portfolio 2021
Materias:
R
Q
Acceso en línea:https://doaj.org/article/16a104e2fdf341d29ce16d4a005364ed
Etiquetas: Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
id oai:doaj.org-article:16a104e2fdf341d29ce16d4a005364ed
record_format dspace
spelling oai:doaj.org-article:16a104e2fdf341d29ce16d4a005364ed2021-11-28T12:16:26ZGeneral intelligence disentangled via a generality metric for natural and artificial intelligence10.1038/s41598-021-01997-72045-2322https://doaj.org/article/16a104e2fdf341d29ce16d4a005364ed2021-11-01T00:00:00Zhttps://doi.org/10.1038/s41598-021-01997-7https://doaj.org/toc/2045-2322Abstract Success in all sorts of situations is the most classical interpretation of general intelligence. Under limited resources, however, the capability of an agent must necessarily be limited too, and generality needs to be understood as comprehensive performance up to a level of difficulty. The degree of generality then refers to the way an agent’s capability is distributed as a function of task difficulty. This dissects the notion of general intelligence into two non-populational measures, generality and capability, which we apply to individuals and groups of humans, other animals and AI systems, on several cognitive and perceptual tests. Our results indicate that generality and capability can decouple at the individual level: very specialised agents can show high capability and vice versa. The metrics also decouple at the population level, and we rarely see diminishing returns in generality for those groups of high capability. We relate the individual measure of generality to traditional notions of general intelligence and cognitive efficiency in humans, collectives, non-human animals and machines. The choice of the difficulty function now plays a prominent role in this new conception of generality, which brings a quantitative tool for shedding light on long-standing questions about the evolution of general intelligence and the evaluation of progress in Artificial General Intelligence.José Hernández-OralloBao Sheng LoeLucy ChekeFernando Martínez-PlumedSeán Ó hÉigeartaighNature PortfolioarticleMedicineRScienceQENScientific Reports, Vol 11, Iss 1, Pp 1-16 (2021)
institution DOAJ
collection DOAJ
language EN
topic Medicine
R
Science
Q
spellingShingle Medicine
R
Science
Q
José Hernández-Orallo
Bao Sheng Loe
Lucy Cheke
Fernando Martínez-Plumed
Seán Ó hÉigeartaigh
General intelligence disentangled via a generality metric for natural and artificial intelligence
description Abstract Success in all sorts of situations is the most classical interpretation of general intelligence. Under limited resources, however, the capability of an agent must necessarily be limited too, and generality needs to be understood as comprehensive performance up to a level of difficulty. The degree of generality then refers to the way an agent’s capability is distributed as a function of task difficulty. This dissects the notion of general intelligence into two non-populational measures, generality and capability, which we apply to individuals and groups of humans, other animals and AI systems, on several cognitive and perceptual tests. Our results indicate that generality and capability can decouple at the individual level: very specialised agents can show high capability and vice versa. The metrics also decouple at the population level, and we rarely see diminishing returns in generality for those groups of high capability. We relate the individual measure of generality to traditional notions of general intelligence and cognitive efficiency in humans, collectives, non-human animals and machines. The choice of the difficulty function now plays a prominent role in this new conception of generality, which brings a quantitative tool for shedding light on long-standing questions about the evolution of general intelligence and the evaluation of progress in Artificial General Intelligence.
format article
author José Hernández-Orallo
Bao Sheng Loe
Lucy Cheke
Fernando Martínez-Plumed
Seán Ó hÉigeartaigh
author_facet José Hernández-Orallo
Bao Sheng Loe
Lucy Cheke
Fernando Martínez-Plumed
Seán Ó hÉigeartaigh
author_sort José Hernández-Orallo
title General intelligence disentangled via a generality metric for natural and artificial intelligence
title_short General intelligence disentangled via a generality metric for natural and artificial intelligence
title_full General intelligence disentangled via a generality metric for natural and artificial intelligence
title_fullStr General intelligence disentangled via a generality metric for natural and artificial intelligence
title_full_unstemmed General intelligence disentangled via a generality metric for natural and artificial intelligence
title_sort general intelligence disentangled via a generality metric for natural and artificial intelligence
publisher Nature Portfolio
publishDate 2021
url https://doaj.org/article/16a104e2fdf341d29ce16d4a005364ed
work_keys_str_mv AT josehernandezorallo generalintelligencedisentangledviaageneralitymetricfornaturalandartificialintelligence
AT baoshengloe generalintelligencedisentangledviaageneralitymetricfornaturalandartificialintelligence
AT lucycheke generalintelligencedisentangledviaageneralitymetricfornaturalandartificialintelligence
AT fernandomartinezplumed generalintelligencedisentangledviaageneralitymetricfornaturalandartificialintelligence
AT seanoheigeartaigh generalintelligencedisentangledviaageneralitymetricfornaturalandartificialintelligence
_version_ 1718408061352148992