Decentralized Intelligent Network under the Wave of AI Revolution

The rapid development of artificial intelligence technology is reshaping our world. Breakthroughs in areas such as deep learning and natural language processing have made AI applications ubiquitous. Since the emergence of ChatGPT in 2022, the AI industry has experienced explosive growth, with a variety of AI tools emerging one after another. It is expected that by 2030, the AI market size will reach $185 billion.

However, the advancement of AI technology has also brought challenges such as data centralization and uneven distribution of computing power. Currently, the AI industry is mainly monopolized by a few tech giants. Against this backdrop, the decentralization concept of Web3 offers new possibilities for addressing these issues.

A batch of high-quality Web3+AI projects has emerged. Among them, Bittensor takes a unique approach by building an AI algorithm platform with a self-filtering mechanism through blockchain competition and incentive mechanisms, providing a new idea for retaining the best AI projects.

Bittensor: Decentralization AI Network

Bittensor is a decentralized machine learning network and digital goods marketplace. It operates on a distributed computer network, addressing issues such as data centralization. Its fair incentive mechanism ensures that subnets and nodes are rewarded based on their contributions. The network provides services to individuals in need of machine learning computing power and becomes a marketplace for trading various types of data.

Bittensor was founded by a group of technology enthusiasts in 2021, using the Substrate framework to build a blockchain. The Alpha version was released in 2022 to validate feasibility, introducing Yuma consensus to protect privacy. The Beta version was released in 2023, introducing TAO tokens to incentivize network maintenance. In 2024, DHT technology will be integrated to improve efficiency and further expand subnets and digital goods markets.

TAO is the native token of Bittensor, with a total supply of 21 million coins, halving every four years. It has been fairly launched with no allocation reserved for the team and investors. Currently, 1 TAO is generated every 12 seconds, with a daily output of approximately 7200 coins, distributed to subnets and their nodes based on contribution.

Currently, the total number of Bittensor accounts exceeds 100,000, with 80,000 non-zero accounts. The price of TAO skyrocketed by dozens of times at one point, with a current market value of 2.278 billion USD and a unit price of 321 USD.

Subnet: The Core Architecture of Bittensor

The Bittensor protocol supports network participants in exchanging machine learning capabilities and predictions, facilitating collaborative model sharing. The network is composed of multiple subnets and employs a survival of the fittest mechanism.

A subnet is an independently operating unit of code that establishes unique incentives and functions while maintaining the same consensus interface as the mainnet. Currently, there are 45 subnets, expected to increase to 64 between May and July 2024.

Subnets include three roles: owners, miners, and validators. Owners are responsible for the underlying code and incentive mechanisms, miners provide computing power and compete with each other, and validators evaluate contributions and receive staking rewards. The subnet emissions are distributed among these three roles in the proportions of 18%, 41%, and 41%.

After the subnet registration, there is a 7-day immunity period, and the first registration fee is 100 TAO. When the capacity is full, the subnet with the lowest emission during the non-immunity period will be deleted. Therefore, the subnet needs to continuously increase its staking amount and efficiency to ensure survival.

Unique Consensus and Proof Mechanism

Bittensor innovatively introduces the Proof of Intelligence ( PoI ) mechanism, combining Yuma consensus to achieve validation and reward distribution:

• PoI mechanism: Miners prove their contribution by completing smart computation tasks, and validators assess and score the quality of the tasks.
• Yuma Consensus: Comprehensive consideration of the validator's staking amount and rating, removing outliers, and distributing rewards. Following the principle of data unawareness to protect privacy.
• MOE mechanism: integrating multiple expert models to work collaboratively, verifiers rank the model capabilities and reward them, promoting optimization.

Popular Subnet Projects

Currently, Bittensor has 45 registered subnets, 40 of which have been named. The more notable ones include:

• Vision 19: Decentralization of image generation and inference, providing access to LLM and image models.
• Cortex.t 18: Build AI platform, providing text and image API services.
• Subnet Generation No. 1: Bittensor's first subnet project.

Other types of subnetworks include data processing, trading AI, and more. Successfully operated subnetworks yield considerable profits, but new nodes require high-performance equipment and optimized algorithms to survive in the competition.

Future Outlook

• Web3+AI will long be a market hotspot.
• Bittensor architecture innovation provides AI teams with a Decentralization migration path.
• Subnet expansion may bring risks such as a decrease in registration thresholds and a decline in earnings.

Overall, Bittensor, with its unique technical architecture and incentive mechanism, provides a new paradigm for the decentralized development of the AI industry, which is worthy of continued attention.