In recent years, blockchain and AI (artificial intelligence) have become two of the most trending and groundbreaking technologies. The idea of combining them offers extremely promising innovations.
This article will give you an overview of both technologies, the benefits and pitfalls of their combination. At the same time, we will also explore a few outstanding projects in this field.
What is Blockchain?
Blockchain is a database system stored distributedly by many members in the network. The information stored in each block is arranged and inseparably linked through encryption algorithms.
What is AI?
AI stands for Artificial Intelligence. It is a model that simulates human intelligence using a computer system. AI can learn and make decisions independently.
The benefits of combining Blockchain and AI
To understand the benefits of combining these two leading technologies together, we will first review their properties.
Properties of Blockchain and AI
A combined system of Blockchain and AI will help it inherit many good properties of both. Here are some possible benefits:
Better data mining capabilities
The ability to exploit data and self-learn is one of the outstanding strengths of Artificial Intelligence, but it is limited by data accessibility. For example, currently ChatGPT cannot access social data sources such as X (Twitter), Facebook directly, but users need to provide it. That's just a small example, there are many data sources that limited AI cannot access.
This can be solved by the Permissionless and Transparent nature of Blockchain. AI freely exploits data without having to ask for a license from a third party. Not to mention, information authenticated on Blockchain is somewhat more secure than other sources thanks to its economic nature.
Make more accurate decisions
As mentioned, Blockchain helps AI freely exploit quality data sources combined with its simulation and prediction capabilities to make better decisions.
On the contrary, because blockchain is transparent, users can easily retrieve the information that AI uses to check whether the decision is trustworthy or not.
Further, if each validator in the blockchain network is operated by Artificial Intelligence, the final decision made will be the decision of the multi-AI system, and thus reliability will be increased. We can imagine a new consensus mechanism, Proof-of-AI, replacing Proof-of-Work or Proof-of-Stake, for example.
Enhance security and privacy
Artificial intelligence has the ability to simulate and predict situations based on real events, which can help the network predict attack situations in advance to make decisions to prevent them in advance.
For users, AI has the potential to increase the privacy of their transactions. Similar to Zero-Knowledge Proof, users can ask AI to perform a transaction or certain action without revealing too much information because AI has the ability to predict and simulate execution.
Increase transaction speed
At first glance, it seems unconvincing because of the current incompatibility of the two technologies. But in the future when they integrate together smoothly, transaction speed will increase significantly.
For example, by using its predictions, AI selects the appropriate group of validators to help transactions go through accurately and quickly. Not to mention, through the process of learning over time, AI can also offer faster and more accurate ways to resolve transactions.
Promote the common development of both industries
Until now, everything inside AI systems has been secure, it is like a black box that is never opened. Blockchainization of AI helps it inherit all the properties of Blockchain, especially "open source".
Not to exaggerate, but "open source" is one of the key things driving Blockchain's development so quickly. In just one night, programmers can clone a hot trending dApp into their own. From then on, the correction continued to develop into better things.
Not to mention the system that combines the two also takes advantage of the crypto-economics - cryptonomics which is the trademark of blockchain. This combination will make the expansion speed of both faster than ever.
Challenges for AI in Blockchain
Performance and speed
AI systems are operated by artificial neural networks (ANN) with hundreds of billions of parameters.
Number of parameters of GPT-3 and GPT-4
To operate effectively, data transmission needs to flow throughout the network at high speed. As you know, the current Blockchain is slow, just a few projects have events and the network is overloaded.
Although expansion solutions are being widely deployed, this is still not enough for AI systems to operate stably.
Ability of extension
Storage and usage costs are also a barrier for AI systems to become truly decentralized. AI data needs to be continuously transmitted, changed and stored to serve its self-learning process. Unfortunately, at the present time, the cost of storing on the blockchain is too expensive, so fully integrating AI into the blockchain will not be economically viable.
Data privacy and security
It is true that if the two technologies are successfully combined, it will bring privacy and increase data security. But before that, we need to find a way to make AI capable of learning and using data on the blockchain without revealing the details of that data.
Because to ensure privacy and security, data posted to the network will usually be encrypted, so if AI wants to use it, it must decrypt or have another method to use when needed and still achieve success. get high speed.
Outstanding projects
Render Network (RNDR)
Render Network is defined as a decentralized GPU rendering platform. This system allows users to contribute GPU (video card) power from their devices to help those who need to render videos, images or 3D effects.
Render Network's operating model is as follows:
- Create job: Through the OctaneRender web interface, customers create a job and submit the file to be rendered (ORBX format). The system relies on file information to calculate costs, time and form of payment through a smart contract. If the customer accepts the terms, they will send RNDR tokens to the smart contract for the work to be done.
- Job assignment: Render Network's MTP protocol will automatically assign the appropriate node operator (GPU sharer) to perform the render job requested from the client.
- Completing work: Node operators use OctaneRender to process assigned work. Once completed, the node operator will send the product to the customer via the Render Network.
- Verify product standards: During work performance, customers can monitor and report on misconduct if necessary. Product previews are blurred until the job is confirmed and paid for by the customer.
- Payment: After agreeing with the product quality, the customer verifies to pay the node operator. Render Network collects a small percentage of RNDR tokens ranging from 0.5% to 5% to support transactions and operate the platform.
Integrating Stable Diffusion - an application that creates images from text using AI and releasing an API that allows applications to easily access the nodes system has made the platform go one step further into the AI segment.
In fact, for Render Network, AI is still just very small integration steps. Computational power is still only isolated in each node, but we have not yet seen a Decentralized AI that uses computing power from the entire network.
Fetch.AI (DONE)
Fetch.AI is an infrastructure layer combining Blockchain and Artificial Intelligence that allows creating a free open economy on top of it.
Fetch.AI is organized into 4 main layers:
User communication tool
Network communication tool
Tool to interact with on-chain ledger
Ledger chain
Ledger - Fetch.AI Chain is built on Cosmos-SDK using Useful Proof of Work consensus mechanism. Fetch chain where smart contracts are built and state stored for members.
Two other important components we need to pay attention to are:
- Autonomous Economic Agents - Autonomous Economic Agents (AEA): are entities registered by users to participate in the system, it can be devices or services that they bring into this economy.
- AEA Framework - Autonomous Economic Agent Framework: also known as OEF (Open Economic Framework) - Open Economic Framework. It is a network connecting multiple agents that allows them to search, explore and interact with each other, thereby allowing them to make their own decisions based on mutual learning.
We can see that Fetch's Machine Learning is integrated into this section, at the AEA Framework layer. This integration helps entities in the network communicate and process tasks with each other automatically and at the same time learn from each other to develop. The results of the work will be transferred and stored on-chain in the Ledger ledger.
In terms of the level of AI integration, Fetch is still only integrated at the off-chain level to handle the above tasks, and the blockchain infrastructure layer still does not have much.
SingularityNET (AGIX)
SingularityNET is a platform that connects AI services through a blockchain network. This platform allows people in need of using AI to find and use the desired service provided.
SingularityNET is implementing 3 main areas:
- AI Marketplace: AI application market, is where customers can search and buy desired services.
- AI Publisher: Portal helps developers manage and publish their AI services.
- AGIX Staking: Staking Portal where users can lock their tokens to secure activity on AI Marketplace and receive rewards.
How SingularityNet works is as follows:
- AI service providers are called Agents, they act as Nodes in SingularityNet.
- Many Agents in the network provide many different AI services, all connected and able to communicate back and forth.
- Agents are ranked after each transaction to discover the advantages and disadvantages of each of them. This is to identify which Agent is good at which job, thereby navigating and dividing the work later.
- When a job arises, the system will automatically search for a suitable Agent to perform and return the results to the user.
- Payment activities take place on-chain through two components: Daemon and Multi-Party Escrow.
The goal that AGIX aims for is to build a general Artificial Intelligence - AGI (Artificial General Intelligence), an Artificial Intelligence system that gathers many artificial intelligences.
Although this solution still cannot be called truly Decentralized AI, it is quite promising if the network connecting them is truly decentralized and each agent in the network must open the black box (open source) for members to use. Other members in the network mine data and learn from each other.
Phase 2 of SingularityNet is trying to solve the speed problem by deploying on Layer 2 and expanding to Cardano. At present, SingularityNET is still in the form of a Decentralized AI Marketplace, which is true to its nature.
Conclude
You see, Blockchain and AI are the two most popular technologies today. Theoretically, the system combining both technologies brings many outstanding benefits, opening up new economic models and solutions with many applications.
But because of their complexity and unique characteristics, this combination is still only superficial, unable to fully promote the good properties of both. Maybe we need to wait more for both to develop to a level enough to reach each other.