Elon Musk famously said that artificial intelligence (AI) is “more dangerous than nukes.” But can blockchain bring more trust and security to AI? Blockchain can track how AI systems make decisions (solving the “black box” problem). And since the blockchain can’t be tampered with, there’s an added layer of security from outside attacks. Guest writer Stevan Mcgrath explores.
Today, we find AI systems in pretty much every technological gadget. It has become so common that it’s difficult to find a system that doesn’t rely on it.
While usually only flashy, groundbreaking uses of AI are reported on, today it is used for basically everything. Things like translations, photo retouching, and even software updates lean on an AI methodology.
But there is, of course, a dangerous side to AI. How much can you trust the decisions made by AI systems? Blockchain, potentially, offers a solution. Let’s start by understanding how blockchain technology works in relation to AI.
How Blockchain and Artificial Intelligence Work Together
As one of the most important advances of the last decade, blockchain works by tracking and recording all transactions about a given market. All its entries are permanent and impossible to edit, making them as trustable as anything in computing can be.
On top of that, the “ledger” is distributed, existing in thousands of computers at the same time. Every time anything is added, the existing data is compared to that in other computers, thus guaranteeing its integrity.
Blockchain’s main use so far is directly tied to cryptocurrency. It was, after all, originally developed as a ledger for all bitcoin-related transactions. All bitcoin data is kept in the Bitcoin blockchain, although it is encrypted for the safety of its users. The most widely used blockchain in the world, the Ethereum blockchain, works in much the same way.
Now, there are other uses for blockchain. There are proposals for its use in a wide array of markets, from real estate to health care. While not all these proposals might come to fruition, blockchain is poised to revolutionize not one, but several markets.
With such a potent tool, it is only natural to look for ways to merge it with anything we can. When it comes to AI, blockchain has shown promise to help begin a new era. Meanwhile, AI itself has been proposed as a tool to make blockchain more trustable.
The Problem with AI…
While AIs are almost ubiquitous today, usually they’re only blindly deployed on low-risk systems. For text prediction on a phone, for example, the risk is minimal and there are no reasons to mistrust an AI.
The same can’t be said about all industries. One of the main issues AIs face on larger scale projects is how difficult trusting their choices can be.
This is partly due to the “black box” nature many AI systems have. “Black box” refers to the problem in which we don’t fully understand how machine learning develops or how AI decisions are made. You can often see an input and a result, but no information about the reasoning is offered. It’s particularly troublesome for systems based on neural networks.
It requires blind trust in the AI system and its decision.
While this is fine for some systems, it isn’t always so. If systems require human input it’s generally fine, as humans will often catch any errors in the computer’s logic. But with completely automated systems, a faulty AI is a liability. A faulty AI can, after all, not just hinder but also at times outright destroy a whole process.
Blockchain Does it Better
Blockchain could help fix this problem. By keeping a secure record of every single action an AI takes, auditing the software becomes much easier. This secure ledger of every decision made by an AI gives users insight into how it is working. By better learning how it works, we could work on making more robust, reliable AI systems.
Blockchain is better for these processes than a regular log simply because it can’t be tampered with. A bank’s software for transactions log could be hacked to delete suspicious transactions, for example. A blockchain record can’t.
By using a permanent, impossible-to-edit record of all steps, systems can be made more secure. A bank could then audit its ledgers knowing all its data is accurate. Or a company could check its optimization processes, knowing everything in it was genuinely done by the AI.
Giving an AI access to a blockchain can also bring huge benefits. In some cases, such as healthcare, blockchains are being proposed to deal with sensitive or private data. While giving a person access to this would be risky, an AI wouldn’t pose a privacy risk. Moreover, it could use the historical data from health care systems to establish patterns.
It could then apply these patterns to patients to predict if, or when, they are likely to suffer health problems. The patients could then receive automated warnings, letting them know visiting the doctor wouldn’t be a bad idea.
AI Can Improve Blockchain, Too
AI and blockchain are a perfect fit. You can think of it as a symbiotic relationship. Both technologies can be used to improve the security and reliability of each other. In other words, AI can make blockchain better, too.
While blockchain is still considered impossible to crack, the same can’t be said about the systems surrounding it, like crypto exchanges. Crypto exchange platforms have been hacked and millions of dollars in bitcoin stolen, thanks to poor security.
Artificial Intelligence could be used to help curb these kinds of attacks. By having an intelligent system learn transaction patterns of its users, it would be possible to flag suspicious ones. These transactions would then be taken to review, as banks often do, and require confirmation from both parties. The key would be finding and establishing use patterns so that only truly strange transactions are flagged.
In summary, while the above is just a pick of possible uses, a greater horizon lies ahead for blockchain and AI. We can be assured that we’ll see a hybrid of these technologies sooner than later. When we do, mankind will be the better for it.