The Intersection of AI and Blockchain

AI and Blockchain: An Introduction

Thanks to advances in machine learning, the pace of development in artificial intelligence (AI) is progressing rapidly. The capabilities of services like ChatGPT, which include writing everything from code to blog posts to music, explaining complex concepts, and brainstorming ideas, are still being discovered. The new technology also offers exciting opportunities to introduce automation, improve quality, and even reduce risks to human life – for instance, by modeling data to predict threats like pandemics.

However, there are also many genuine concerns about the pace of advancement as few people understand how the algorithms work on a granular level. Further, the data used to train algorithms can be subject to human biases, resulting in unintended outcomes such as discrimination against minority groups.

The fact that algorithms are mostly owned and controlled by centralized companies rather than being open-source makes them even more opaque and, critics argue, prone to misuse for profit.

Blockchain is generally based on open-source technology, where code and transaction data are publicly available and, thus, subject to scrutiny. Anyone can join the network or run applications on it, and any user can interact with decentralized applications (dapps) using only a crypto wallet.

Blockchains also are not typically owned or operated centrally and are subject to decentralized governance, meaning no central entity can take control and corrupt the network for its own ends.

As such, the features of blockchain could offer the potential to overcome some of the challenges facing AI development.

Benefits and Challenges of Combining AI and Blockchain

There are several benefits that can be anticipated from combining AI and blockchain, although they aren’t all easily achievable.

Increased transparency

AI is said to have an ‘explainability’ problem, meaning that even the designers of an algorithm often can’t explain precisely how it reaches its results, often due to the sheer quantity of data that AI algorithms process. If the outcome can’t be explained, the technology may be deemed untrustworthy. For example, if there’s an inability to account for events, such as how an accident occurred where a driverless vehicle was at fault, who would carry the blame?

Making the algorithm transparently available via the blockchain ledger introduces transparency and accountability to the underlying code. This would allow anyone to scrutinize it to understand how it reached a particular outcome. Further, if AI data and algorithms are made available on open-source marketplaces based on blockchain, it could help to democratize access to them and introduce transparency around their use.

However, implementing ‘explainability’ via blockchain in practice will likely require centralized entities to relinquish control over their existing intellectual property and make it open-source, which is a significant barrier by itself.

Lastly, the ability to explain AI and its code will not necessarily overcome a loss of trust in the event that the technology is deemed responsible for causing a harmful incident.

Automation

Combining AI and blockchain-based smart contracts introduces new opportunities for automation and reducing human intervention in repetitive or high-frequency tasks.

One example could be using AI to develop quantitative trading strategies and implementing them using smart contracts that trigger based on specific market events.

Data storage and computing capacity

AI algorithms require a vast amount of computing capacity and data to run, and the latter also requires storage. Growing this infrastructure to meet demand on a centralized model will consume resources such as real estate and hardware.

Decentralized infrastructure projects such as Filecoin or Storj can crowdsource these resources from large networks of individual and enterprise users willing to lend their idle resources. This approach could reduce the climate impact of the growing demand for AI.

Ethical Considerations for AI and Blockchain

While there are clear ethical benefits to increasing the transparency and accountability of AI, there are also some ethical pitfalls to consider.

The public, transparent nature of blockchain data must be balanced against individual data privacy. The benefits of achieving a permanent, immutable record of data cannot outweigh the legal right to be forgotten under regulations such as the EU’s General Data Protection Regulation (GDPR).

Decentralization can help to promote equality and democracy, but it also comes with ethical considerations when applied to AI. For example, there is no central authority to take responsibility if an open-source algorithm produces unintended outcomes that cause harm, limiting the chance for restitution.

Achieving truly decentralized governance at scale comes with many practical challenges, such as organizing votes and ensuring voter participation, which may act as a barrier to the effective governance of AI.

Use Cases for AI and Blockchain Integration

Currently, one of the most popular use cases for combining AI and blockchain is in generative art. Typically, a digital artist will create a master theme for a collection of art and allow an algorithm to generate a potentially infinite number of pieces, each with unique traits created by the algorithm. Each one is then attached to an NFT, conferring digital scarcity. Famous examples include Bored Ape Yacht Club and Cryptopunks.

DeFi and crypto investing offer another use case for AI. For example, companies such as TokenMetrics use AI to analyze investing trends, with their AI system trained to analyze on-chain data to support investment decisions.  

AI analysis of on-chain data is also used by firms such as CertiK to examine large volumes of blockchain data for cybersecurity and fraud detection purposes, as well as auditing smart contracts.

Healthcare is another area of opportunity. Healthtech firm Vytalyx enables healthcare professionals to use AI to provide customized medial and dietary advice, with access and outcomes recorded on the blockchain.

There are also several blockchain projects that incorporate AI, such as SingularityNET,  a decentralized marketplace for AI services, allowing AI developers and companies to monetize their data and algorithms. Other examples include Numerai, a quantitative hedge fund that crowdsources machine learning models to attempt to predict the stock market, and Fetch.ai, a network of AI autonomous agents that can carry out tasks (like booking a flight).

The Future of AI and Blockchain Development

The use cases and initiatives listed here are only a few areas of potential opportunity in converging AI and blockchain. Since both technologies are relatively new and each is still being developed, there could be huge scope to apply them across a variety of sectors in different ways. However, since there are practical and ethical issues to overcome, the success of any given initiative is not guaranteed.  

There are also potential regulatory hurdles to overcome. Blockchain regulation is a contentious topic with a patchwork of laws across the globe, while AI is even less regulated. However, given the financial and societal implications of both technologies, a future involving regulated use seems inevitable.

Blockchain and AI essentials

• The features of blockchain, such as transparency, permissionless access, and an immutable trail of transactions, offer significant potential to help reduce the opacity and explainability of AI algorithms.

• Smart contracts can also support AI automation, while decentralized infrastructure could help to reduce the environmental impact of meeting demand for AI.

• There is significant potential for development at the convergence of blockchain and AI, with many use cases under exploration; however, there are practical, ethical, and regulatory hurdles to overcome.