Introduction

Decentralization in the realm of blockchain technology has induced a radical shift in our perception of data security, transparency, and reliability. But the evolution awaits yet another paradigm shift with a fusion of A.I. and blockchain in the form of Oraichain.

Oraichain with its AI powered Layer 1 blockchain aims to introduce A.I. to every possible aspect of decentralized Applications (dApp). Oraichain bridges the power of AI computation with smart contracts in a qualitative and trustworthy way. It has an AI Layer that incorporates a variety of services and applications including AI Oracles. In this article, we will explore the Orachian infrastructure a decentralized AI (DeAI) network for building AI powered dApps and SaaS projects on-chain.

Mainnet

The Oraichain mainnet is built on top of Cosmos SDK with IBC interoperability. It is secured by a robust Proof-of-Stake (PoS) validator network. It supports CosmWasm smart contracts and offers zero-knowledge cryptographic verification for AI proofs and test cases. To know more about the components of the Oraichain mainnet, explore the Github repositories.

Architecture

In the following sections, we will understand about the Oraichain AI Oracle service. To begin with, we’ll first go through all the entities involved individually and then go through their interplay as a system to act as the AI Layer.

Key Terms AI Provider

Oraichain allows anyone to become an "AI Provider" by offering AI-powered data sources. When these data sources are used, the providers earn rewards. The reward rate is clearly stated when the data source is created. The reward system incentivizes these AI Providers for the creation and maintenance of high-quality AI data sources for the Oraichain ecosystem.
<aside>🖼️ INFOGRAPHIC REQUIRED - All aspects under an entity symbol
</aside>

AI Data Source

Oraichain allows AI providers to create "AI Data Sources" by registering their AI models on the network. The metadata of the Data Source is lives on the chain for transparency and contains information such as the owner, name, transaction fee, source code, and description.
The main aim of their existence is to give a sustainable and effective solution to provide data for AI model development in a decentralized way. By data contribution from multiple independent sources, the potential biasness and noise induced in data collection can be mitigated.

Oracle Script

The Oracle Scripts are used to fulfil data requests on the Oraichain Oracle Service. These are the files that are responsible for collecting data from various AI Data Sources or other Oracle Scripts. They use user-provided test cases to evaluate this data. If the data satisfies the test cases, then the script initiates an aggregation phase. The aggregated data is then stored on the Oraichain for future use and verification.
An Oracle script on Oraichain stores the following parameters on the chain:

• Name of the Oracle script

• Owner of the Oracle script

• Overview of the Oracle script

• Minimum fees is required to run this Oracle script. This is an approximation based on the fees set by the data source and test case providers

• A List of Data Sources used in the script

• A List of Test Cases used in the script
  
  

  

AI Test-case

Similar to every software development cycle where testing is an essential aspect of delivering reliable software. Test cases are also fundamental to Oraichain’s A.I. ecosystem. A test case contains encrypted inputs and outputs that help the Oracle Scripts examine the reliability of the data source.
A test case, as defined by the owner must contain the following components which are stored on-chain:

• Owner of the test case

• Name of the test case

• Txn Fees for executing the test case

• And finally, the overview of the test case

Note - Since the actual code might be large, it is advised to store it off-chain.

AI Testcase Provider

By including Test Case Providers as a key aspect of the ecosystem, Oraichain focuses on blockchain entities that provide AI test cases that help ensure the validity and quality assurance of the results. Like the AI Providers, test case providers also receive proportional rewards for their services.

Infrastructure

Now, let’s explore the trustworthy Layer 1 AI infrastructure. For a detailed understanding, we will delve into it individually.

Trustworthiness

Trust is one of the key aspects of any decentralized network. Similarly, for the Oraichain network, it’s crucial to become trustworthy i.e. the results from our AI executors is not determined by one single entity but is rather dependent on multiple entities.
And, furthermore it is non-biased in nature while being easy-to-measure in an automatic or deterministic way.

We can delegate the overall trustworthiness for A.I. into the following components -

1. Testcase Data - Reliable testcase data in itself leads to reliable and trustworthy results from the AI Execution

2. Performance Evaluation Tools - By evaluating performance of all the entities present in the Oracles Service by factors like responsiveness (opposite of down time) , we can rank the performance of the entities to better understand their nature

3. Intent & Behavior Prediction Tools - These tools can predict the potential behavior of AI service providers based on past data and performances. This aids in maintaining a trustworthy environment by identifying and managing potential risks or malicious intentions.

Layer 1 A.I.

The current system of Execution follows a schedule as depicted in the infographic.

Overview
Oracle Chain is a blockchain-based platform that provides a secure and transparent way to perform AI API execution. It uses a decentralized network of validators to ensure that tests are performed accurately and efficiently.
Process
The following steps are followed from the start of the request and up to the end when the results are received.

1. A user creates an AI Request on the blockchain

2. The request is routed to a random validator from the pool of available validators, who are then tasked to run according to the Oracle Script and start fetch the data from the AI API.

3. The validators then run the provided testcases or the ones fetched from testcase providers to check the reliability of the data source.

4. If found correct, then the validators pass on the results back to the user as the Response.

Trustworthy Proofs

As we briefly mentioned in the last section about the role trustworthiness in A.I. execution. Now we’ll utilize that knowledge to understand the **Trustworthy Proofs** that live on the intersection of Zero Knowledge (Zk) and A.I.

To understand what determines trustworthiness, we can divide it into several factors and understand the individually to gather the bigger picture-

1. Accuracy - It  refers to the degree to which the output of an AI model matches the expected result. High accuracy indicates that the AI model is reliable and can be trusted to make accurate predictions or decisions.

2. Verifiability - It is key to the execution of AI that the outcomes of the model can be independently verified. This ensures a good general confidence score in the utilized model.

3. Integrity - The integrity of an AI model refers to its ability to perform consistently under multitude of conditions.

4. Reliability - This relates to the model's capability to give consistent behavior and results over time.

5. Safety - It refers to the capability to perform without causing any intentional or erroneous results.

6. Security - The general inexistence of vulnerabilities help build trust in the execution.

7. Speed - As with many realtime applications, low latency is very crucial. Thus, speed of execution matters in A.I. execution as well.

The most essential aspect of Trustworthiness is to measure or analyze it in a deterministic way, by utilizing zk-proofs to its benefit and other major techniques too. Orai aims to obtain trustworthiness in a decentralized way. Some of the major components utilized in doing so are-

• Randomized Testcases

• Zero-Knowledge Proofs

• Decentralized Data Storage

• Use of Virtual Machines

• Serialization

Decentralization Of A.I.

Now, to utilize all this amazing decentralized A.I. architecture, several platforms have been established and are under development by the Oraichain Team. Here we’ll discuss two of those that play a key role in bringing all the things discussed above to the real world.

Data Hub

Data Hub is a fully decentralized an monetized platform that aims to provide functionalities around exploring, storing, sharing, creating or requesting high-quality datasets that suit real-world A.I. use-cases.
ℹ️ Data Hub

AI Hub

Orai Chain's AI Marketplace is decentralized marketplace platform that enables anyone to offer their curated A.I. services in the form of packages. By the power of A.I. Oracles these can then be linked to your smart contracts for utilization in real-world DApps. This enables seamless integration of A.I. services into DApps with a fair incentivization to the service creator.
ℹ️ AI Hub

Conclusion

Oraichain is at the fore front of revolutionizing the modern world with a fusion of Blockchain and Artificial Intelligence. By using A.I.’s derivative powers to creatively generate insights and understand complex patterns from large sets of real-time data. Orai focuses on bringing this paradigm shifting use-case to the ever changing landscape of blockchain. Leveraging this into privacy focused, decentralized and transparent real-world applications can empower developers and businesses alike. And drive innovative, efficient and secure solutions for growth in both the blockchain and A.I. space.