Blockchain Data Management Benefits by Increasing Confidence in Datasets Supporting Artificial Intelligence (AI) and Analytical Tools using Supply Chain Examples

Abstract

We describe how Hyperledger Fabric (HLF) blockchain (BC) technology that we previously applied to Navy logistics supply chains can be applied to data supporting artificial intelligence (AI) and software development in terms of system safety and the timely acquisition of data. Data-driven AI/machine learning (ML) requires trusted data for their use in AI functions and requires significant amounts of training data from diverse sources including Internet of Things (IoT) devices/sensors. Unauthorized alterations to data supporting AI/ML could go unnoticed within the AI function build process but surface during operation in hazards affecting unwanted human death or resource destruction. AI/ML controlling hardware usually falls into the two highest software control categories: Levels 1 and 2, risk of death, disability, or resource destroyed. HLF BC is a tamper-resistant decentralized trusted ledger that provides proof of transaction where trust is implemented through distributed consensus to ensure that only authorized people can modify data and that the modification is traceable and transparent. Distributed ledgers provide system safety through BC provenance, immutability, and policy enforcement through smart contracts. We show how BC can contribute to the safety of the data and transactions and provide data to the researchers in a timely manner through “smart repositories.”