Gold-Standard AGI: 1: Normative ASI Superalignment

{"In":[0],"order":[1],"to":[2,101,110,136,149],"maximise":[3],"the":[4,102,111,117,138,150],"net":[5,126],"benefit":[6],"of":[7,43,104,128,142,154],"AGI":[8,35,113],"(Artificial":[9],"General":[10],"Intelligence,":[11],"and,":[12],"in":[13,97,116],"particular,":[14],"agentic":[15,146],"superintelligent":[16,122],"AGI,":[17,95],"a.k.a.":[18],"Artificial":[19],"Superintelligence,":[20],"or":[21],"ASI)":[22],"for":[23,94],"all":[24],"humanity,":[25],"without":[26],"favouring":[27],"any":[28],"subset":[29],"thereof,":[30],"we":[31,57,75],"imagine":[32],"a":[33,59,77,89,98,144],"Gold-Standard":[34],"that":[36,81,119],"is":[37,121,135,172],"practically-maximally-aligned":[38,145],"and":[39,70,91,170],"practically-maximally-validated.":[40],"The":[41,125],"first":[42],"these":[44],"properties":[45],"---":[46,48],"alignment":[47,54,72,114],"may":[49],"be":[50],"decomposed":[51],"into":[52],"normative":[53,112],"(how":[55,73],"do":[56,74],"define":[58],"final":[60,167],"goal":[61,168],"$\\\\mathbf{FG}_G$":[62,83],"which,":[63],"if":[64],"pursued":[65],"as":[66,84],"intended,":[67],"ensures":[68],"practical-maximal-alignment?),":[69],"technical":[71],"build":[76],"practically-maximally-validated":[78],"agent":[79],"$G$":[80,120],"pursues":[82],"intended?).":[85],"This":[86],"paper":[87],"presents":[88],"conceptual":[90],"philosophical":[92],"foundation":[93],"culminating":[96],"proposed":[99,130],"(and,":[100],"best":[103],"our":[105,129],"knowledge,":[106],"novel)":[107],"implementation-neutral":[108],"solution":[109,131],"problem":[115,141,153],"case":[118],"(hence":[123],"\\"superalignment\\").":[124],"effect":[127],"(the":[132],"$\\\\mathbf{TTQ}$+$\\\\mathbf{NAP}$":[133],"combination)":[134],"reduce":[137],"(seemingly":[139],"impossible)":[140],"building":[143,155],"superintelligence":[147],"$S$":[148,163,171],"(much":[151],"easier)":[152],"an":[156],"$\\\\mathbf{NAP}$-compliant":[157],"non-agentic":[158],"\\"goal-less":[159],"superintelligent-agent-under-construction\\"":[160],"$S^-$":[161,165],"(where":[162],"=":[164],"+":[166],"$\\\\mathbf{TTQ}$,":[169],"practically-maximally-validated).":[173]}

Authors

Institutions

Publication Details

Journal
Zenodo (CERN European Organization for Nuclear Research)
Published
2026-09-14
DOI
https://doi.org/10.5281/zenodo.22743647
Primary Topic
Computability, Logic, AI Algorithms
Type
preprint
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
preprint

Gold-Standard AGI: 1: Normative ASI Superalignment

Aaron Turner
Zenodo (CERN European Organization for Nuclear Research)
Computability, Logic, AI Algorithms
preprint

Gold-Standard AGI: 1: Normative ASI Superalignment

Aaron Turner
preprint en

Abstract

In order to maximise the net benefit of AGI (Artificial General Intelligence, and, in particular, agentic superintelligent AGI, a.k.a. Artificial Superintelligence, or ASI) for all humanity, without favouring any subset thereof, we imagine a Gold-Standard AGI that is practically-maximally-aligned and practically-maximally-validated. The first of these properties --- alignment --- may be decomposed into normative alignment (how do we define a final goal $\mathbf{FG}_G$ which, if pursued as intended, ensures practical-maximal-alignment?), and technical alignment (how do we build a practically-maximally-validated agent $G$ that pursues $\mathbf{FG}_G$ as intended?). This paper presents a conceptual and philosophical foundation for AGI, culminating in a proposed (and, to the best of our knowledge, novel) implementation-neutral solution to the normative AGI alignment problem in the case that $G$ is superintelligent (hence "superalignment"). The net effect of our proposed solution (the $\mathbf{TTQ}$+$\mathbf{NAP}$ combination) is to reduce the (seemingly impossible) problem of building a practically-maximally-aligned agentic superintelligence $S$ to the (much easier) problem of building an $\mathbf{NAP}$-compliant non-agentic "goal-less superintelligent-agent-under-construction" $S^-$ (where $S$ = $S^-$ + final goal $\mathbf{TTQ}$, and $S$ is practically-maximally-validated).

Zenodo (CERN European Organization for Nuclear Research)
Laboratori Guglielmo Marconi (Italy) (IT)
Computability, Logic, AI Algorithms
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.