Promoting Distributed Trust in Machine Learning and Computational Simulation

Bore, Nelson; Raman, Ravi Kiran; Markus, Isaac M.; Remy, Sekou L.; Bent, Oliver; Hind, Michael; Pissadaki, Eleftheria Kyriaki; Srivastava, Biplav; Vaculín, Roman; Varshney, Kush R.; Weldemariam, Komminist

doi:10.1109/bloc.2019.8751423

Cited by 11 publications

(15 citation statements)

References 13 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Before self-improvement technology can be unleashed, AI behavior control systems need to be developed and tested in transparent, non-hackable simulation sandbox environments as proposed by Bore et al [49] seems essential.…”

Section: Discussionmentioning

confidence: 99%

“…Smart contract terms could require 2/3 or 100% acceptance of DLT-authenticated (Axiom 6) signatories to ASILOMAR AI Principles or similar regulatory documents. Smart contract terms can deny access to those who do not fulfill a transparency requirement via Supplier's Declaration of Conformity [32], which document could in turn require inclusion of an accepted set of ethics and morality (Axioms, 2, 3) and a safety testing record meeting certain standards [18,49], all of which can be incorporated into a CI (Axiom 5). Equally critical, dApps permit separation and balance of powers of key AGI components, analogous to no one entity having all the nuclear launch codes.…”

Section: Decentralized Applications (Dapps)mentioning

confidence: 99%

“…The IBM Research Supplier's Declaration of Conformity via a factsheet for AI software incorporates an audit trail as a fundamental principle [32]. Bore et al describe a system for incorporating an audit trail in DLT as part of embedding AI simulations in DLT so that trust in the simulations' validity is enabled between researchers without requiring a trusted intermediary [49].…”

Section: Audit Trail Of Component Usage Stored Via Distributed Ledgermentioning

confidence: 99%

“…To address #1 and #2, we need a control system that is effective enough and transparent enough to prevent those paths, such as through Axioms 2 and 3, transparent and unhackable ethics and morality, and Axiom 4, the behavior tree value system. Bore et al take the goal of transparent simulation and modeling to a new level by describing a system wherein simulation specifications and an audit trail are stored via DLT, thus facilitating a means to cross-validate simulations before deployment and obstruct malicious hacking or fraud in simulations by humans or AI [49]. Sandbox problem #3 may be redressed with the separation and balance of powers described next.…”

Section: The Need For Simulation Of Control and Value Alignmentmentioning

confidence: 99%

See 3 more Smart Citations

Safe Artificial General Intelligence via Distributed Ledger Technology

Carlson¹

2019

Preprint

View full text Add to dashboard Cite

Artificial general intelligence (AGI) progression metrics indicate AGI will occur within decades. No proof exists that AGI will benefit humans and not harm or eliminate humans. I propose a set of logically distinct conceptual components that are necessary and sufficient to 1) ensure various AGI scenarios will not harm humanity and 2) robustly align AGI and human values and goals. By systematically addressing pathways to malevolent AI we can induce the methods/axioms required to redress them. Distributed ledger technology (DLT, ‘blockchain’) is integral to this proposal, e.g. ‘smart contracts’ are necessary to address evolution of AI that will be too fast for human monitoring and intervention. The proposed axioms: 1) Access to technology by market license. 2) Transparent ethics embodied in DLT. 3) Morality encrypted via DLT. 4) Behavior control structure with values at roots. 5) Individual bar-code identification of critical components. 6) Configuration Item (from business continuity/disaster recovery planning). 7) Identity verification secured via DLT. 8) ‘Smart’ automated contracts based on DLT. 9) Decentralized applications - AI software modules encrypted via DLT. 10) Audit trail of component usage stored via DLT. 11) Social ostracism (denial of resources) augmented by DLT petitions. 12) Game theory and mechanism design.

show abstract

Section: Discussionmentioning

confidence: 99%

Section: Decentralized Applications (Dapps)mentioning

confidence: 99%

Section: Audit Trail Of Component Usage Stored Via Distributed Ledgermentioning

confidence: 99%

Section: The Need For Simulation Of Control and Value Alignmentmentioning

confidence: 99%

See 2 more Smart Citations

Safe Artificial General Intelligence via Distributed Ledger Technology

Carlson¹

2019

Preprint

View full text Add to dashboard Cite

show abstract

“…Smart contract terms could require 2/3 or 100% acceptance of DLT-authenticated (Axiom 6) signatories to ASILOMAR AI Principles or similar regulatory documents. Smart contract terms can deny access to those who do not fulfill a transparency requirement via Supplier's Declaration of Conformity [33], which document could in turn require inclusion of an accepted set of ethics and morality (Axioms 2,3) and a safety testing record meeting certain standards [11,52], all of which can be incorporated into a CI (Axiom 5). Equally critical, dApps permit separation and balance of powers of key AGI components, analogous to no one entity having all the nuclear launch codes.…”

Section: Decentralized Applications (Dapps)mentioning

confidence: 99%

Safe Artificial General Intelligence via Distributed Ledger Technology

Carlson

2019

BDCC

View full text Add to dashboard Cite

Artificial general intelligence (AGI) progression metrics indicate AGI will occur within decades. No proof exists that AGI will benefit humans and not harm or eliminate humans. A set of logically distinct conceptual components is proposed that are necessary and sufficient to (1) ensure various AGI scenarios will not harm humanity, and (2) robustly align AGI and human values and goals. By systematically addressing pathways to malevolent AI we can induce the methods/axioms required to redress them. Distributed ledger technology (DLT, “blockchain”) is integral to this proposal, e.g., “smart contracts” are necessary to address the evolution of AI that will be too fast for human monitoring and intervention. The proposed axioms: (1) Access to technology by market license. (2) Transparent ethics embodied in DLT. (3) Morality encrypted via DLT. (4) Behavior control structure with values at roots. (5) Individual bar-code identification of critical components. (6) Configuration Item (from business continuity/disaster recovery planning). (7) Identity verification secured via DLT. (8) “Smart” automated contracts based on DLT. (9) Decentralized applications—AI software modules encrypted via DLT. (10) Audit trail of component usage stored via DLT. (11) Social ostracism (denial of resources) augmented by DLT petitions. (12) Game theory and mechanism design.

show abstract

Information-Theoretic Approaches to Blockchain Scalability

Raman

Varshney

2022

Springer Optimization and Its Applications

View full text Add to dashboard Cite

Promoting Distributed Trust in Machine Learning and Computational Simulation

Cited by 11 publications

References 13 publications

Safe Artificial General Intelligence via Distributed Ledger Technology

Safe Artificial General Intelligence via Distributed Ledger Technology

Safe Artificial General Intelligence via Distributed Ledger Technology

Information-Theoretic Approaches to Blockchain Scalability

Contact Info

Product

Resources

About