Blockchain + AI, Decentralised Artificial Intelligence: The start of a new Era

Blockchain and AI are one of the most powerful technologies currently in the market, merging both can result in super-effective technology

Introduction

AI and Blockchain are two technologies that are currently booming in the tech space. We have seen the great advancements that are happening in Blockchain and AI space, but have we ever thought, or worked on how these two spaces when merged act like. It will be a tech boom and yes it will empower the technology space that we currently have. Let me start with a short idea of the two technologies and then discuss how these two can help each other and create a positive impact on how we use these technologies at present.

About AI

Artificial Intelligence as you might be knowing is making technology think and act. We can code a lot of stuff but when it comes to automatic predicting and learning of things, that is where AI helps you. Artificial Intelligence comprises data, machine learning models, deep learning models, and reinforcement learning. We give data to a model which after analyzing and processing it, learns how its data points behave and then we can use these models for predicting future events. Mathematically we say we want to find the best possible m (Which is basically weights) in this equation y = mx+c where y and x are given (Say in 1000’s lakhs, which is what we call data).

About Blockchain

Blockchain on the other hand is a public ledger or you can say a decentralised database where whatever data you upload on it doesn’t reside on a central server instead it remains on nodes/computers that are connected to this network. The key features of blockchain are: It is immutable (You can’t alter on-chain data) It is transparent (Everything is clean and clearly visible on a public network) It is secure, and authentic (With the help of some advanced cryptography algorithms) The identity of the user is hidden behind a public address

How Blockchain is helping AI

You can see the data present on the Blockchain is publicly available. Currently when we work on AI models some of the problems that we face are about:

• Authenticity of data, where it is coming from
• Is it a biased dataset
• Gathering and uploading data
• Privacy issues as you are sharing data with third parties
• The infrastructure required to process and build AI Models

All the upper issues discussed are the key ones and impact a lot of startups and projects that want to build projects that contain key user identity details which in turn hampers the growth in this domain.

Blockchain on the other hand as already discussed is available on the public ledger and the identity of users is behind a public address. This solves one of the biggest problems we face nowadays!

Let’s discuss one by one how the upper problems can be solved by blockchain:

• The authenticity of data, the problem is easily solvable as the whole data present on the blockchain is signed by someone. It is authentic and has someone behind that. You can’t manipulate on-chain data as it is not possible on a blockchain network due to its immutable feature.
• Biased is one set of problems we face in AI; we can’t say that it can be removed completely in any sense by blockchains but yes, the bias due to manipulation and non-availability of data can be minimized
• Gathering and uploading of data and other infrastructure requirements can be greatly dealt with by blockchains as we discussed the data is freely available in the public domain, and for those who want to hide data for them, private blockchains are the key. Regarding infrastructure the whole blockchain network is a cluster of computers, we can make a public Artificial Intelligence blockchain network where people can share their infrastructure resources and get paid in return in the form of tokens. This is actually already a feature of blockchain networks we have but for bigger processing needs a separate network can be useful. In fact, IBM is working on one such project where they are creating a decentralized Artificial Intelligence system.
• Also, the overhead of uploading and gathering data can be reduced as data is freely available on chain and we can easily take data in real-time which can greatly improve the current AI space
• Privacy Issue is already solved due to the anonymous feature due to the public address system
• Another great feature would be that we can see code on which data is being processed or predicted instead of a model/code that is present behind a cloud server. This will provide genuineness of prediction.
• We can also track changes happening to the AI model, data, etc . and incentivize the contributors accordingly.

In short AI models needs data and a large dataset which to be recent one helps in developing better models. Blockchain as it is a real time trustworthy data sharing network, it helps to solve all of these problems. Microsoft is also working a kind of similar model, Infact they might incentivize people for improving and adding relevant data to it in form of tokens which is a key use case of blockchain networks. They are also trying to make a kind of system which can run on everyday devices and apps.

These were some key features that can be solved in AI domain, there are a lot many. We can make a public Artificial Intelligence model on Ethereum, and allow people to modify and contribute to it. Research and Development is already Happening in this space. Soon we will see a decentralised AI which empowers business’s and economies.

AI helping Blockchain

Till now we discussed how blockchain can empower AI, Now let’s understand how AI can help a lot of dapps that are currently in the market and how it can empower financial markets of blockchain where AI has not been reached as of now.

Crypto assets are a kind of strong financial data that might be very difficult for humans to intercept and understand. AI can prove a great help in such cases. AI can be used for pattern detection and predictive algorithms for trading assets, tokenizing funds and improving auditability.

AI can also help in reducing transaction costs and automatic submitting of transactions based on network analysis. Also, a key role in which machine learning can come into play is in the making of stablecoins. Stablecoins basically are those crypto tokens whose value doesn’t fluctuate a lot and mostly remain fixed. At present based on some algorithms and managing liquidity through contracts we try to stabilize the token value aka price but sometimes due to many issues and the way we liquidate like based on the current time we fail to stabilize the coin and it fluctuates to higher level. AI model if deployed to control the price can predict the upcoming price of a token and then can trigger functions that can add or remove liquidity before time. This can help in making the real stablecoin we intend to form on the blockchain.

Apart from this AI can be used to build agents which can trade on-chain digital assets increasing profits and helping customers to earn. As blockchain is something transparent customers can freely trust it and can review it in times of need. Also, it provides a better view of how efficient and practical that model would be, making it easy for customers to find the right set of applications to use.

An example

I was sometime back working on a project where I was using user’s social media data to predict their credit score but the problem I faced while making a one was non availability of social media data of users on a public domain. I would have to resort to asking for the same from the companies and that was also quite impossible to do due to privacy concerns on their end. There were dummy data but it was not trustworthy and it might not be the real reflection of what real data looks like. I did not need personal information but still, you know due to privacy concerns such data can’t be made public. At that time, I thought of decentralized social media clubbing with my credit scoring platform. As data is decentralized, we can use Realtime data. It is trustworthy and the privacy of the user is also hidden. For the Image analysis part, we can use a decentralized AI model that takes data and predicts it on the user side itself without relieving the real image. You can see how this can empower the machine-learning projects we want to make.

Conclusion

At last, I would like to say AI and Blockchain are quite different technologies but the mix of both actually fills the needs of each other. Making AI and Blockchain go hand in hand can provide better models, better blockchain applications, and better wider use cases of these technologies. Many tech giants and startups are working in this area and I am hoping for a great set of inventions in this domain!

Thanks, I hope you liked it!