Edit Question Tags: Why wouldn't AGI's utility function change over time like with humans?

From Stampy's Wiki

You do not have permission to edit this page, for the following reason:

We are migrating to Google Docs / coda. Please hold onto your edits for now.

The action you have requested is limited to users in the group: Administrators.

---

Tags:

AcademiaAcausal tradeActorsAgencyAgent foundationsAgiAgi fire alarmAiAi safety campAi safety supportAi takeoffAi takeoverAlgorithmic progressAlignmentAlignment proposalsAlignment targetsAnthropicArcArchitecturesArms raceAsiAutomationAutonomous weaponsAwarenessBenefitsBiological cognitive enhancementBoxingBrainBrain-computer interfacesCapabilitiesCareersCenter for human compatible ai (chai)CivilizationClrCognitive enhancementCognitive superpowersCoherent extrapolated volitionCollaborationCommunicationCommunityComplexity of valueComprehensive ai servicesComputeComputing overhangConjectureConsciousnessContentContributingControl problemCooperative inverse reinforcement learning (cirl)CorrigibilityCounterfactualsCreativityCybersecurityCyborgismDebateDeceptionDeceptive alignmentDecision theoryDeep learningDeepmindDefinitionsDeveloppersDifferential technological developmentDifficulty of alignmentDo what i meanDoomEducationEliezer yudkowskyElkElon muskEmbedded agencyEmotionsEnculturedEpistomologyEric drexlerEthicsEutopiaEvolutionExistential riskFarFormal proofFriendly aiFundingFuture of humanity instituteGame theoryGoalsGoodhart's lawGovernanceGovernmentGptGpt-3HedoniumHuman valuesHuman-in-the-loopHumansImpact measuresImplementationIncentivesInfohazardInformation securityInfra-bayesianismInfrastructureInner alignmentInside viewInstitutionsInstrumental convergenceIntelligenceIntelligence amplificationIntelligence explosionInterpretabilityInvestmentingLanguage modelsLiteratureMachine learningMaximizersMegaprojectsMentorshipMesa-optimizationMetaethicsMetaphilosophyMetaphorsMiriMolecular nanotechnologyMotivationMultipolarMyopiaNarrow aiNatural languageNearest unblocked strategyNeeds workNeural networksNeuromorphic aiNick bostromNoneObjectionsOpen problemOpenaiOrganizationsOrthogonality thesisOther causesOughtOutdatedOuter alignmentOutreachPaperclip maximizerParadigmPattern-matchingPeoplePerson-affecting viewPersonal actionPersuasionPhilosophyPhysicsPlausibilityPlex's answer to what are some good resources on ai alignment?PolicyPoliticsPower seekingPreferencesProductivityProgrammingPsychologyQuantilizersRace dynamicsRay kurzweilRecursive self-improvementRedwood researchRegulationResearch agendasResearch assistantsResourcesRobin hansonRobotsRobustnessS-riskScaling lawsSecretSecurity mindsetSeed aiShard theorySimulation hypothesisSingletonSolutionsSpecification gamingStable win conditionStampyStop buttonStubStudySuccess modelsSuperintelligenceSurveysTech companiesTechnological unemploymentTechnologyTestTest tagTimelinesTool aiToolingTransformative aiTranshumanismTreacherous turnTripwireTruthful aiUtility functionsValue learningValuesWhat aboutWhole brain emulationWhy not justWireheading

Cancel

Retrieved from "https://stampy.ai/wiki/Special:FormEdit/Question_Tags/Why_wouldn%27t_AGI%27s_utility_function_change_over_time_like_with_humans%3F"