Next Steps In The Integration Of Artificial Intelligence And The Blockchain

Having worked in the cryptography space for over two decades, and having been an active participant in the cryptocurrency evolution since its inception, I take a deep interest in the subject. In particular, I believe that the intersection of artificial intelligence (AI) and blockchain is an exciting but challenging new development.

Matt Turck recently discussed why the topic matters and highlighted interesting projects in the space, referring to AI (big data, data science, machine learning) and blockchain (decentralized infrastructure) as the defining technologies of the next decade. Evidently, the time is already ripe for these new concepts, despite them being novel and still underdeveloped.

Intriguingly, AI and blockchain are philosophically different in various ways:

  1. AI is driven by more centralized infrastructures as opposed to blockchain’s decentralized, distributed nature.
  2. While a lot of AI technologies are owned and operated by centralized providers, a majority of the blockchain players in the market publish all of their codebases as open-source code that is freely available for anyone to inspect at any point in time.
  3. AI is more of a black-box solution for now, while the blockchain tends to be more transparent in all the transactions processed.
  4. AI is based on probabilistic formulas, while blockchain is more deterministic in nature.

Currently, AI startups are being overwhelmingly acquired by companies such as IBM, Apple, Facebook, Amazon, Google, Intel and Alibaba, among others. These organizations rely on unprecedented amounts of data to train their AI agents, which offers them an immense competitive advantage. At the same time, their data and capabilities are closed from the rest of the world.

Unfortunately, centralized AI introduces room for abuse, such as massive surveillance of people using face recognition and computer-vision-powered technology. At the same time, creating solutions on top of a centralized environment requires enterprises to give up privacy and control of their data to other third parties.

Merging AI And Blockchain 

This is where blockchain comes in, as it can be used to overcome many of AI’s shortcomings. I’ve seen it firsthand in our business, where we leverage a lot of AI and machine learning (ML) capabilities in order to better identify and authenticate users’ blockchain identities.

Currently, experts in this space are exploring ways through which blockchain can be deployed to create a decentralized marketplace to enhance AI. This course by MIT is just one indicator of the movement in this space. This will allow people to comfortably share their personally identifiable information (PII) with the assurance that it will remain secure and private through decentralization and secure computing offered by the blockchain. In effect, users can easily share their sensitive details, such as health and financial data, and the system would ensure that only the intended service provider would have the ability to decipher and decrypt users’ PII with explicit consent by the user. With time, I believe the space will have an accumulation of massive data maintained by big organizations using AI algorithms to stay competitive.

An article on Hackernoon lists some of the latest projects integrating blockchain and AI technologies to create cutting-edge solutions. Some of the more notable ones include SingularityNET, an AI marketplace where enterprises can acquire AI capabilities on a global scale to enhance the growth of the space. Another project is Namahe AI, a platform that aims to improve the efficiency of supply chains by integrating AI and blockchain to enable seamless monitoring of the processes in real time and flagging anomalies and fraud for review. Finally, there’s Numerai, an AI-based hedge fund that sponsors competitions for industry enthusiasts to develop and submit prediction models and solutions.

Challenges Of Merging AI And Blockchain 

Obviously, AI solutions differ from legacy ones, since they follow probabilistic models. In other words, a traditional program follows the approach of “IF A happens, THEN follow B.” On the contrary, AI (deep learning and machine learning) uses probabilistic answers to follow a succeeding step. This feature of AI makes the technology ideal for creating flexible solutions. Nevertheless, the tradeoff is that some AI programs make mistakes.

To date, AI agents still go wrong in some cases, and it still remains difficult for users to know when it is wrong or what should be done when it makes a mistake. A few memorable examples include a Microsoft chatbot gone rogue, Wikipedia edit bots engaging in feuds among themselves, Uber’s self-driving cars ignoring red lights and Russian robot Promobot IR77 escaping the laboratory.

Another issue is compliance. It is still a major concern to control AI solutions from going rogue or causing damage. AI and blockchain solutions will require data aggregation, which is a real challenge. However, the internet of things (IoT) will be vital in the provision of data required for AI training. In effect, the security and privacy of privately owned data will be crucial in this space.

Talent is another challenge for merging blockchain and AI. While data, which is the primary factor for training AI models, can be gathered using IoT devices, professionals will be needed to develop algorithms that run in a decentralized or distributed manner as required in blockchain technology. Fortunately, organizations such as Deep Brain Chain and SingularityNET are continuously researching and creating innovative AI algorithms.

Computing resources still remain an issue in merging AI and blockchain. Luckily, it is possible to leverage global idle computing power to run resource-intensive AI training integrated with blockchain.


Some experts are now suggesting blockchain has the capability of decentralizing AI to achieve decentralized intelligence available to the masses. To reap the real benefits from their integration, I believe it will be imperative to address several major concerns: how to determine when an AI solution is wrong in its operations, how to train professionals in the field and the need to come up with appropriate compliance requirements to guide the development and deployment of the products. To make real progress, today’s players in the field should seek to break down these roadblocks and encourage blockchain and AI development in the real world.

Crypto Destroyer