AI Revolution: How Bittensor Reshapes Collective Intelligence Networks

The Explosion of the AI Industry and Its Integration with Web3

The rapid development of artificial intelligence technology is leading us into a whole new data-driven era. Breakthrough advancements in fields such as deep learning and natural language processing have caused AI applications to flourish everywhere. The emergence of ChatGPT in 2022 completely ignited the AI industry, followed by an explosive emergence of various AI tools, from text generation to video production, AI is permeating all industries. It is expected that by 2030, the market size of the AI industry will reach $185 billion.

However, the current development pattern of the AI industry is still dominated by a few tech giants, which inevitably leads to issues such as data monopoly and uneven distribution of computing power. At the same time, the decentralized concept of Web3 offers new possibilities for addressing these challenges. Under the distributed network architecture of Web3, the development pattern of AI is expected to be reshaped.

At the intersection of AI and Web3, a number of innovative projects have emerged. Fetch.ai leverages blockchain to build a decentralized AI economy; Numerai uses blockchain and a community of data scientists to predict markets; Velas creates a high-performance AI + blockchain smart contract platform. However, most of these projects focus on a single application scenario, making it difficult to form a complete ecosystem.

Bittensor has keenly seized this opportunity by building an AI algorithm platform with a built-in competitive filtering mechanism through blockchain's incentive system, providing a good ecological environment for high-quality AI projects. This innovative attempt is expected to reshape the development pattern of the AI industry.

Bittensor: The Pioneer of Decentralized AI Networks

Bittensor is a decentralized incentivized machine learning network and digital goods marketplace. It has the following notable features:

• Decentralized Architecture: The Bittensor network consists of thousands of distributed nodes, effectively addressing issues such as data centralization.

• Fair incentive mechanism: The rewards of the network to the subnet are proportional to its contributions, and the reward distribution within the subnet also follows the same principle.

• Open machine learning resources: Provide services for individuals in need of machine learning computing power.

• Diversified digital goods market: not only supports AI model trading, but can also be expanded to various types of data trading.

The development history of Bittensor fully reflects its characteristics as a "geek project":

• In 2021, initiated by a group of technical experts, a foundational blockchain was built using the Substrate framework.

• In 2022, the Alpha version of the network was released to validate the feasibility of decentralized AI. Yuma consensus was introduced to protect user privacy.

• In 2023, the Beta version will be launched, introducing the TAO token incentive mechanism.

• In 2024, apply DHT technology to enhance data processing efficiency, with a focus on developing subnetworks and digital goods markets.

The token economic model of Bittensor is quite unique. The total supply of TAO tokens is 21 million, halving every four years, and it is issued fairly with no pre-mining or team allocation. Currently, one block is produced every 12 seconds, with each block rewarding 1 TAO. Approximately 7200 TAO are produced daily, distributed among various subnetworks and their internal nodes based on contribution.

As of now, the Bittensor network has over 100,000 accounts, with 80,000 active accounts. In the past year, the price of TAO has increased by more than several dozen times, with a current market value of approximately $2.278 billion and a price of $321. These data fully demonstrate the development momentum and market recognition of Bittensor.

Subnet Architecture: The Core Innovation of Bittensor

Subnets are the most critical components of the Bittensor network architecture. Each subnet can be seen as a piece of independently running code, equipped with specific incentive mechanisms and functions, while maintaining a consensus interface consistent with the main network. Currently, the Bittensor network has a total of 45 subnets, and it is expected to increase to 64 by mid-2024.

The subnet mainly includes three types of roles:

• Subnet owner: responsible for providing the underlying code and designing the incentive mechanism.

• Miners: Responsible for iterating and optimizing services and mining code to stay ahead in the competition.

• Validator: responsible for assessing subnet contributions and ensuring their correctness.

The emission mechanism of the subnet ( is the core of reward distribution in the Bittensor network. Generally, 18% of the rewards obtained by the subnet are allocated to the owners, 41% to the validators, and 41% to the miners. The subnet adopts a survival of the fittest mechanism, where poorly performing nodes will be eliminated.

After the subnet registration, there is a 7-day immunity period. The initial registration fee is 100 TAO, and the price doubles for subsequent registrations. When all subnet slots are filled, the subnet with the lowest emission rate that is not in the immunity period will be replaced by a new subnet. Therefore, subnets need to continuously improve the validator staking amount and miner efficiency to ensure long-term survival.

The subnet architecture of Bittensor provides a solid foundation for the implementation of decentralized AI data networks. For example, Masa, as the first dual-token reward system, has successfully landed in the Bittensor network and secured $18 million in funding.

![Bittensor: How AI Subnet Reshapes Collective Intelligence Networks?])https://img-cdn.gateio.im/webp-social/moments-1c84109da5ff6ffbd817edf590d0eec0.webp(

Innovative Consensus and Proof Mechanism

The Bittensor network employs various innovative consensus and proof mechanisms, among which the most distinctive are the Proof of Intelligence ) PoI ( mechanism and the Yuma consensus.

The PoI mechanism requires miners to prove their contributions by completing intelligent computing tasks. These tasks may involve natural language processing, data analysis, or machine learning model training. Validators are responsible for assigning tasks and scoring them based on the quality of completion.

Yuma consensus is the core consensus algorithm of Bittensor. When aggregating validator scores, it assigns different weights based on the amount of TAO staked by validators and eliminates results that deviate significantly from the majority opinion. Ultimately, rewards are distributed based on the comprehensive score.

An important feature of the Yuma consensus is the "data obliviousness principle", which allows nodes to perform computation and verification without needing to know the specific content of the data being processed, effectively protecting user privacy.

In addition, Bittensor also introduces the MOE) mixture of experts( mechanism, integrating multiple expert-level sub-models within a single model architecture. This allows expert models from different fields to collaborate and produce better results than a single model. Validators can score and rank the expert models and allocate corresponding rewards, thereby incentivizing the models to continuously optimize.

![Bittensor: How AI Subnet Reshapes Collective Intelligence Networks?])https://img-cdn.gateio.im/webp-social/moments-bb4bc6663bb70a74645896ab5b06ee64.webp(

Subnet Project Overview

Currently, the Bittensor network has 45 registered subnets, of which 40 have been named. As the number of subnets increases, the registration difficulty has decreased, but in the long run, it will still be a process of survival of the fittest.

Apart from the root subnet, the most attention is currently on subnet 19, 18, and 1, whose emission proportions are 8.72%, 6.47%, and 4.16%, respectively.

Subnetwork 19, Vision, focuses on decentralized image generation and inference, providing access to various image models. Currently, its nodes have an average daily income of approximately 866 USD.

Subnet 18 Cortex.t is dedicated to building a cutting-edge AI platform, providing text and image services through APIs. Its nodes have an average daily yield of about $554.

Subnet 1 is the earliest text generation subnet. Although it has faced skepticism, it still maintains a high ranking.

In addition, there are some specialized subnets, such as Meta Search) analyzing social media data to provide market sentiment( and Omron) optimizing staking strategies with neural networks(, etc.

Overall, successfully operating subnet nodes can yield considerable profits, but continuous optimization is also required to survive in intense competition.

![Bittensor: How AI Subnet is Reshaping Collective Intelligence Networks?])https://img-cdn.gateio.im/webp-social/moments-a6491289020557c0f4df9c6f4fd1a48f.webp(

Future Outlook

1. The integration of Web3 and AI will continue to be a market focus, attracting significant investments.

2. Bittensor, as a non-traditional VC project, combines technological innovation and market recognition, and has a good development prospect.

3. Its innovative subnet architecture lowers the barrier for AI projects to migrate to decentralized networks, which is expected to attract more high-quality projects to join.

4. As the number of subnets increases, the TAO rewards obtained from the original subnet may decrease, requiring the TAO price to rise accordingly to maintain the level of returns.

5. Lowering the registration threshold for subnets may bring some low-quality projects, but in the long run, the competitive mechanism will ensure that high-quality projects prevail.

![Bittensor: How AI Subnets are Reshaping Collective Intelligence Networks?])https://img-cdn.gateio.im/webp-social/moments-a6a0a9cd30f27b7e81c269677cfe6de7.webp(