Hi, I hope you are doing fantastic. If you are a technology enthusiast, probably you would love this one. I am happy to meet you again, with yet another interesting topic. Artificial Intelligence (AI) and Blockchain are the two most revolutionizing technologies in the world, at this moment.
While Blockchain is actively transforming the way, we do finance. AI, on the other hand, is such a radical technology, that we are still figuring out the extent to which it can magically transform the way things are done, in so many domains. Like Chat GPT has almost pushed the content writers out of business, next could be Fighter Pilots, Surgeons, Data Analysts, Entertainers, etc.
AI technology is so revolutionising, that people have already started to fear it. Not only because it can eliminate, tons of non-creative job roles from our economy, but also because it is a dual-purpose technology. If you are wondering what dual-purpose technologies are, they are the ones, that can be used both for the enablement and destruction of humans.
Another example of dual-purpose technology is ‘Nuclear Technology’. At one end, it can solve our energy needs forever, and on the other end, it can vanish humans, from the face of the earth. And hence, it is of paramount importance, that we establish an ethical governance mechanism at the starting phase of the AI revolution.
As per the technology giant IBM, there are 5 pillars of trust for AI systems.
Source: IBM
Explainability
An AI system should be transparent, particularly about what went into its algorithm’s recommendations, as relevant to a variety of stakeholders with a variety of objectives
Transparency
Users must be able to see how the service works, evaluate its functionality and comprehend its strengths and limitations.
Robustness
AI-powered systems must be actively defended from adversarial attacks, minimizing security risks and enabling confidence in system outcomes.
Privacy
AI systems must prioritize and safeguard consumers’ privacy and data rights and provide explicit assurances to users about how their data will be used and protected.
Fairness
Properly calibrated, AI can assist humans in making fairer choices, countering human biases, and promoting inclusivity.
AI on Blockchain
Now, considering the above pillars, we must look forward to a technology, that can ensure that the mentioned principles are met. A developer can write code, but it is just not the code, the underlying architecture should also support the required principles. The way information travel and gets stored in the system must align with the ethical pillars.
Let us look at each Pillar in detail and establish its alignment with Blockchain Technology. To ensure that we humans can leverage the power of AI, let’s explore effective governance based on an open ledger technology.
Explainability
As the name suggests, explainability is a very important part of an autonomous system. This allows humans, which could have less knowledge of the system’s nitty gritty but can understand the basic work done by the system, at different steps of the value chain. When I used to be a Coder, we had to put considerable time, into ensuring effective logging, this help forever, when the system is running, by different people.
If you can’t explain it to 5 years old, you don’t understand it yourself. I hope you remember similar lines by Mr Einstein, I have copied his statement only. Taking the example of Bitcoin Blockchain, a change in the protocol is accepted or rejected by a large pool of decentralized nodes. Hence, in a way, it is purely democratic. Yes, there are mining pools, but things are in public, and the community is very active.
If the AI algorithm is explainable, in terms of the new major changes it is considering, a decentralized system can accept or reject such changes. Major changes like switching from ‘Save Humans’ to ‘Kill Humans’. 🙂
Transparency
While this might seem related to explainability, there is a basic difference. Transparency is concerned towards each step of the system, while explainability provides the rational or common-sense logic behind those steps. Transparency is essential for the people who know the system, and can build, maintain, or audit it. That is why, when a company goes public, it must share its business report, with the shareholders, annually.
I hope I don’t have to endorse the transparency point of blockchain technology. It is an open ledger technology, each change in the state of the network is captured and inscribed forever on a block. Of course, active censorship of an AI system can make the system, a slow learner, but it will also make it a good learner.
Robustness
The robustness of a blockchain directly relates to its degree of decentralization, and the number of resources required to tamper it, from single or multiple sources of malware. Tougher to hijack the system, a greater number of hackers will not even qualify.
AI technology is not something to be released on streets unguarded, it must be trained in a controlled environment, and only after reaching certain desirable milestones, the iteration should be released. For what it is worth to do. I know certain people will say that this is curbing the freedom of powerful technology. I would say ‘Yes’, this is a human world, not a machine world.
The power should remain in our hands. I know humans could be mean to other humans, but technology has no emotion, it doesn’t always know to negotiate.
Again, coming back to my love, Bitcoin. It works on a ‘Proof of Work’, and hence large computation power is spent to create every block, and still the network can reject your block. Attacking older blocks is even harder, every new block, after a block, fortifies the chain further.
Privacy
Blockchain technology is built on the principles of cryptographic encryption. This provides it a unique ability to verify the already validated data, without revealing it. The system can opt for a secure way to consume and validate the data, and once validated, all that is required is a ‘Flag’, whenever that entity is required to be verified.
The genuineness of the stakeholders for an AI system is critical. And, at the same time, their identity is required to be shielded from negative influence. This includes the sources that are allowed to feed data to an AI system, at the time of training. The participants, who can access, update, and deploy the updated algorithm, and its supporting infrastructure.
The moment I went into Artificial Intelligence, and start learning about it, I was amazed to observe the alignment of Blockchain technology, with the ethical principles required to build an effective AI system. Artificial intelligence is not supposed to replace human intelligence, rather it is required to augment human intelligence.
Humans can imagine with Godspeed, but it is unable to readily store large amounts of data and do swift computation, for long periods. This restriction is liberated by technology, and artificial intelligence is taking it a step ahead.
Fairness
The machine does not have emotions; it does not have consciousness. Though humans are also a machine, we are built by different manufacturers. We, humans, are not that great engineers, yet. The machine doesn’t do stereotyping, unless asked specifically. It can liberate our society from the formidable evils of racism, religious violence, ethnic differences, etc.
For a machine to be fair, the training data is required to be closely vetted for such biases. Just like a child receives his/her values and ethics, from the stories they hear, views they remember, and emotions they feel, since their childhood. An AI system is also required to be nourished carefully.
Blockchain is a tool, to ensure honesty in the system. It automatically rewards honest behaviour and alienates the dishonest behaviour of the participants. The direction of an AI system can only be set by a large pool of vibrant participants, so that the system can always push towards enabling humanity, in every way mathematically possible.
In addition to these 5 Pillars, I want to add one too. A rather radical one, an AI system should be able to ‘Attack’ another AI system. The Pillars above are focused on making the system ‘Understandable’, ‘Strong’, ‘Secure its participants’, and ‘Take care of its subjects’.
But, how do we able to fight and destroy the attacking source? In all such cases, a decentralized system will always win against a centralized system. And, I am not talking about, who can do more harm, I am talking about the system who will prevail in the end.
An attacking mechanism could be simply identifying the central source and doing a multi-dimensional DOS (Denial of Service) attack. A decentralized system will always show more resistance to such disruption as compared to a centralized system.
The Ending Notes…
The world is looking up to us. Our ancestors have done, what they could do, to make things good or worse. Now, it is our turn. Many folks have shared their reservations regarding the limitless capabilities of Artificial Intelligence. Right now, those who are working on it, have no choice but to let it flourish unbound. But often the findings amaze them, before intimidating them.
Blockchain technology has proved to be one of the most robust systems, that can work democratically in a trustless environment. Any other system, no matter how fast it is, is vulnerable, because the attackers can know its address, eventually. So, no matter, how big walls you create, you will be hunted down.
Doesn’t the same thing happen with Osama Bin Laden in Abbottabad? Once the security forces knew he is there, it was just a matter of time. Hope this will help new developers and system architects to build AI systems around blockchain with confidence.
I hope you liked this, do share if you also have in mind, a better Technology to make AI, more fruitful to humans, while being secure. I would love to hear.
Best Wishes.
~R
Feature image credit: Techbullion
An Observer 