Hi all,

Have you read something about AI recently? It's not like you could avoid the topic – and it is set to become more and more important in the coming months and years.

So what’s up?

AI offers no shortage of opportunities, but it also has some serious issues (and we're not talking about robots that may take over the world). For starters, AI is a centralising force in an already highly concentrated tech landscape.

Turns out we might need blockchains to counter that. Today, we explain why and how.

Let’s get started! 🤓

Artificial intelligence (AI) has a lot of potential – and some important challenges. Even if you don't know much about the technology, you have probably heard about discussions on how to regulate the industry.

Arguably the key advantage of AI is that it can simulate human reasoning, only much faster. That allows for quick reactions based on incredibly large volumes of inputs. At the same time, this is also a challenge. AI decision-making is not only complex, it's also intransparent. 

Complex AI models are essentially a black box. That has significant implications. Can you trust AI? Are AI-based decisions ethical? Which standards are they based on? Neither of these questions has an easy answer.

Can Blockchain solve AI's issues?

Public blockchains have a couple of key features which could balance AI's shortcomings. AI is usually centralised and opaque. Blockchains, on the other hand, are transparent digital records. They can also provide decentralised data storage.

In a nutshell, enhancing AI with blockchain technology could improve trust, privacy and accountability.

Trust and accountability

The lack of AI accountability directly affects trust in the outputs. Any trustworthy AI, however, needs to be built on explainable algorithms.

Using a blockchain which provides an immutable – and transparent – digital record may lead to a better understanding of a specific AI model. That could increase trust in the integrity of the data as well as in recommendations and other AI outputs.

Data integrity and security

Blockchains allow for decentralised storage of data. On the user side, that enables audits about how certain data is actually used. (By the way: Auditing data usage is not limited to AI models.)

When an AI model itself is stored and distributed on a blockchain, the output becomes more decentralised and transparent. In short, it becomes more accountable – although that may not be the main ambition for an AI provider. Regulation as well as market pressure, however, may create a demand for accountability.

Augmentation and automation

Blockchains can help AI to expand by providing access to data, improving management of data consumption, and creating the possibility to share AI models. Ideally, the data market would become more transparent and therefore more trustworthy than today.

Combining AI and blockchain could also reduce the need for human intervention in many types of business processes. Blockchain-based applications can already be used to replace unnecessary third parties. (In many cases, that is theoretical for now. In practice, there are often technical and regulatory stumbling blocks).

When AI is used to automate routine tasks, transparency around the respective AI model is no longer nice to have. Businesses which have to comply with a range of legal requirements cannot simply rely on a black box to process their data.

Computing power

AI needs a lot of IT resources in the background. Centralised servers may not be enough to cope with the demand, especially when certain peaks have to be covered. Blockchains can be used to provide fair access to computing power anywhere in the world.

And how can AI help blockchains?

Blockchain technology can benefit from the impressive development of AI. For example, intelligence captured within blockchain-based corporate networks can be enhanced by AI, giving the respective companies a definite competitive advantage. After all, AI is much better than humans when it comes to reading and analysing large amounts of data while also identifying relevant links.

AI models may also be incorporated into the so-called smart contracts. These applications are blockchain-based but have not been incredibly smart so far. With AI, additional opportunities appear. One very small example: predetermined thresholds and events can trigger payments or re-orders to ensure that companies don't run out of stock.

Moreover, AI is already in use to help detecting fraudulent activity on blockchains. This is another area that has significant potential for growth.

Summary

AI can process huge amounts of data and still react quickly. While this may simulate and enhance human reasoning, there is also a major drawback: AI-made decisions are often not transparent – and may even be influenced by hallucinations.

Core features of blockchains could help to provide insights into opaque AI models. Blockchain technology is based on immutable and transparent records of digital transactions. It also enables decentralised data storage. In combination, these features could boost trustworthiness and accountability of AI models.

You want to learn more about AI? Then you can find a good guide here, but there is no shortage of other websites to cover all angles. For day-to-day developments, you can also sign up to the free newsletter of today's sponsor: The Rundown AI.

That’s the end for today! 😢