Technology Intersections: I Bring the Brains, You Provide the Trust
When Artificial Intelligence Meets Blockchain
In the open-source market segmentation initiative — DLT Landscape — I dedicated a layer to technology intersections that provide powerful value propositions, by combining distributed ledger technology (DLT) — e.g. blockchain — with other established or emerging technologies.
In this post, I’m exploring what happens when artificial intelligence (AI) and blockchain technology meet. It is an exciting topic that comes up increasingly often in industry conversations.
Fighting Digital Disinformation
Tackling digital disinformation is a use case that has attracted the attention of a number of research organizations and technology firms, that have been exploring potential responses to this problem by combining AI capabilities with blockchain technology.
Ironically enough, while AI technology is being used to create manipulated content such as deepfakes, it can also be applied to fight disinformation, by devising it to detect anomalies and inconsistencies in text, images or videos. Adding blockchain into the mix, provides a decentralized and trusted mechanism for verifying the source and other critical metadata for digital content.
I know of a handful of tech companies — e.g. Attestiv and Truepic — that have developed and launched offerings that combine these two technologies to fight digital fraud. The Attestiv platform, for example, validates digital media that is captured by a mobile or web application or an API, or is imported from an edge device or another external source. The captured items are forensically scanned to detect anomalies, and a unique fingerprint is stored on a tamper-proof distributed ledger. The platform makes use of AI for functions such as image categorization, text extraction, and image or video authentication.
Beyond validating an image or a video, distributed ledger technology can also be used to record every step of data processing and decision-making for an AI algorithm, so we are able to trace back and understand what data points led to a particular outcome.
Explainable AI — simply put — is about building and using AI algorithms and models in a way that they both empower and fairly affect individuals and organizations.
The need for explainability has been a topic of debate when it comes to the application of AI, especially in highly-regulated industries such as healthcare. In the healthcare industry in particular, the legal implications of using AI are significant, and there is a constant conflict between innovation and regulation, which needs careful consideration.
Key to explainable AI is to take an open and transparent approach. It is not just about having high-quality data that feeds an algorithm or model, but also about providing trust and transparency throughout the entire lifecycle — essentially, it includes both the understanding of the provenance of the data on which AI models are trained, as well as the detailed recording of these models’ decision-making process — a promising opportunity at the intersection of AI and DLT, to improve the trust and confidence in the data that AI algorithms use, as well as the decisions and outcomes derived from those algorithms.
Decentralized AI Marketplaces
AI research today is mostly dominated by a handful of big techs, and most existing AI marketplaces — e.g. modelplace.ai — which are meant to facilitate access to AI tools and services, are centralized.
After doing some research, I’ve found one example of a decentralized marketplace — currently in beta — where independent developers can publish and monetize AI agents, algorithms, and models. On the SingularityNET Marketplace — originally built on Ethereum and migrated to Cardano in 2021 — developers can launch their AI/machine learning tools, interoperate with other tools, and sell them to users. SingularityNET describes itself as an open and decentralized network of AI services made accessible through blockchain, where independent developers can publish their services and charge users — anyone with an internet connection — using the AGIX digital token.
A somewhat different example is Nokia’s blockchain-backed data marketplace — launched in 2021 — that enables data to be securely exchanged and used for AI and machine learning. It is a B2B play, and was built on the permissioned Hyperledger Fabric distributed ledger protocol. Reportedly, the platform also enables federated learning, a collaborative feature that allows ecosystem participants to collectively train AI/ML models across distributed datasets.
In decentralized marketplaces, the underlying blockchain or distributed ledger provides the necessary trust, smart contracts are used to automate and formalize the relationship between the ecosystem participants, and digital tokens can be utilized for payments. In the case of AI models in particular, democratizing access to and providing the ability to monetize AI models to independent developers, fosters healthy competition and encourages a broader adoption of AI. Additionally, enabling federated learning as a feature is pretty useful as it helps create better AI models, tools, and services.
AI and machine learning have the potential to bring a new level of intelligence to blockchain systems, and enable more informed, actionable insights, which result in smarter decision-making.
Commodity trading, in particular, is an interesting area where blockchain can join forces with AI to boost supply chain sustainability as well as fair trade. In markets where price had traditionally been the only real differentiator, actionable data about the commodity not only can increase production efficiency but also creates opportunities for additional differentiation, such as fair trade, social inclusion and environmental compliance.
Agtech startup Bext360, for example, applies AI capabilities to measure and grade the quality of commodities, as well as predict weather conditions and growing patterns, and uses a distributed ledger to record provenance information and shipment specifications of the commodity, which also serves as a payment ledger to ensure that farmers are paid fairly and promptly.
More broadly speaking, supply chains can become smarter by embedding AI models into smart contracts that are executed on a blockchain or distributed ledger. This could allow for practical gains such as automated recalls of expired products or execution of reorders based on specific events, as well as prompt, automated resolution of disputes.
AI and blockchain on their own are fascinating technologies, however, if we combine them, the resulting value propositions can be even more powerful. Nevertheless, we need to make sure that we use them the right way and for the right reasons.