Blockchain + AI: Combining Technologies for Advanced Capabilities

JL Marechaux
Technoesis
Published in
4 min readJan 4, 2019

What have in common Golden Retrievers, blockchain, the beef stroganoff recipe, and artificial intelligence? All of them have been very popular search topics on Google in the past 2 years .

(Source: Google Trends — January 2019)

In this blog post, I will stay away from the dog breed and the delicious Russian dish to focus more on blockchain and AI. Both have been incredibly popular in the IT industry recently, for all sort of good and bad reasons. Blockchain is often referred to as the next internet revolution because it could potentially change the way we do business. The promise of artificial intelligence and machine learning is to drastically improve the way we understand data and to accelerate decision making. Using your preferred search engine, you can find millions of entries on blockchain, and billions on AI. But surprisingly, there is limited information on blockchain and AI combined.

Nevertheless, if AI is about the automation of cognitive processes and blockchain the automation of transactions, there are probably business needs to leverage both of them simultaneously. The use of machine learning, deep learning and blockchain to solve business problems is relatively new. We can expect that technologies in these areas will keep evolving in 2019 and beyond. We can also envision that for some specific needs, AI and blockchain could be used together. The first section describes some use cases where a blockchain could leverage AI. The second section is about scenarios where an AI solution may benefit from some blockchain capabilities.

AI-powered blockchain

(Using AI as an underlying technologies in blockchain platforms)

Simply put, a blockchain is a peer-to-peer network based on a shared distributed ledger, with self-executing business rules (smart contract), and where information is validated through a consensus mechanism. The key features of a blockchain are not related to artificial intelligence, but AI and machine learning could potentially contribute to bring blockchain to another level.

Improved consensus mechanisms

In a blockchain, each new transaction is collectively verified based on a predefined agreement between participants. There are many different consensus protocols, and some of them, such as the proof-of-work (PoW), are known for using massive amount of energy because of the computing power needed to support them. We can imagine that AI will eventually improve the efficiency of consensus mechanisms, to incrementally replaced brute-force processing approached by more elegant and energy-efficient solutions.

AI could be leveraged to learn how to verify transactions more efficiently in a blockchain network.

Blockchain fraud detection with AI

Blockchain frameworks rely on robust mechanisms to guarantee the security and the privacy of the data it contains. And one core concept in blockchain is to guarantee the immutability of stored information.

As blockchain systems will become more important in the next few years, there is no doubt that specialized hackers and malicious participants will try to access protected data. Given the nature of a blockchain, it may be very difficult for humans to detect malicious behaviors. But AI and ML are very good to detect attack patterns. With supervised learning and pattern recognition algorithms, it is possible to train models on specific blockchain transactions is order to learn about the characteristics of a fraudulent activity.

AI could become very useful in fraud detection and security breach prevention on blockchain networks.

Smarter smart contracts

In essence, a smart contract is a business rule deployed on a blockchain network to execute a set of predefined actions. The value of a smart contracts resides in automation and consistency. As the business rule is scripted, it can be automatically applied when a transaction happens. Moreover, because a smart contract is deployed on every nodes through the blockchain network, it is guaranteed that the same rule will be consistently applied to everyone.

Now what if the business rule could evolve to adapt to changing data and to learn from past events. This is in a nutshell the promise of predictive and prescriptive analytics. If a system is able to learn from historical data (machine learning), then it can improve business rules over time. A smart contract can leverage the results from a prediction model (AI) to refine the business logic of a subsystem.

Machine learning could make the smart contracts smarter.

Now that we have talked about AI as an underlying technology in blockchain platforms, the next story will explore how blockchain can be leveraged for advanced AI use-cases.

Read part 2 here, it will only takes 6 additional minutes of your time.

--

--

JL Marechaux
Technoesis

Data Science & AI/ML at Google. My team is building advanced analytics and applied AI/ML models for large Google customers.