AI Revolution: The Shift from Centralization to Decentralization

The rapid development of artificial intelligence technology is leading us into a brand new data-driven era. Breakthrough advancements in fields such as deep learning and natural language processing have made AI applications ubiquitous. The emergence of ChatGPT in 2022 ignited the entire AI industry, followed by a surge of various AI tools, from text generation to video production, AI is changing the way we work and live. It is expected that by 2030, the market size of the AI industry will reach an astonishing $185 billion.

However, the AI industry is currently mainly monopolized by a few tech giants, which inevitably brings a series of challenges such as data centralization and uneven distribution of computing resources. At the same time, the decentralization concept of Web3 provides new possibilities for addressing these issues. In the distributed network of Web3, the development landscape of AI is expected to be reshaped.

In this context, a batch of high-quality Web3+AI projects has emerged. Among them, the Bittensor project has seized a critical gap - it has built an AI algorithm platform with a built-in screening competition mechanism through blockchain competition and incentive mechanisms, aiming to retain the highest quality AI projects. This innovative model is expected to break the barriers of traditional AI development and bring new vitality to the industry.

Bittensor: Pioneer of Decentralization AI Network

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

• Decentralization: Bittensor operates on a network composed of thousands of distributed computers, effectively addressing issues such as data centralization.

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

• Machine learning resources: Providing services for individuals who need machine learning computational resources.

• Diversified digital goods market: Initially focused on machine learning models and related data transactions, it has now evolved into a platform for trading any form of data.

The development of Bittensor is filled with highlights of technological innovation and community building.

• 2021: Project launched, building blockchain through Substrate framework.
• 2022: Released the Alpha version of the network, introducing Yuma consensus.
• 2023: Release of the Beta version and the introduction of the token economic model TAO.
• 2024: Introduction of DHT technology to improve data storage and retrieval efficiency.

It is worth mentioning that Bittensor has relied less on traditional venture capital during its development and has instead attracted nodes and miners through token incentives, maintaining the network's Decentralization characteristic. Essentially, Bittensor is an AI computing power and service project driven by GPU miners.

The native token of Bittensor, TAO, is similar to Bitcoin in many ways, with a total supply of 21 million coins, and it halves every four years. TAO is distributed through a fair launch method, with no pre-mining or team retention. Currently, the network generates a block approximately every 12 seconds, with each block rewarding 1 TAO. These rewards are distributed to various subnetworks based on contribution, and are then allocated to owners, validators, and miners by the subnetwork.

TAO tokens can be used to purchase computing resources, data, and AI models on the Bittensor network, and they also serve as a certificate for participating in community governance. Currently, the total number of accounts on the Bittensor network has exceeded 100,000, with the number of non-zero accounts reaching 80,000. In the past year, the price of TAO has surged significantly, and its current market value has reached $2.278 billion, with a coin price of $321.

The Subnet Architecture of Bittensor: The Cornerstone of Collective Intelligence

The Bittensor protocol is a Decentralization machine learning protocol that supports the exchange of machine learning capabilities and predictions among network participants, facilitating collaborative sharing of models and services. The protocol includes components such as network architecture, sub-tensors, subnet architecture, as well as validator nodes and miner nodes.

Subnets are the most critical component of the Bittensor network architecture. Each subnet is a segment of independently running code that establishes specific user incentives and functions while maintaining the same consensus interface as the main network. Currently, there are 45 subnets in total, excluding the root subnet. It is expected that by May to July 2024, the number of subnets will increase from 32 to 64, with 4 new subnets added each week.

The subnet mainly includes three roles:

1. Subnet Owner: Responsible for providing the underlying code, setting up incentive mechanisms, and allocating miner work incentives.

2. Miners: Responsible for iterating and optimizing servers and mining code to maintain a competitive edge. Underperforming miners can be replaced by new miners. It is worth noting that miners can run nodes across multiple subnetworks.

3. Validator: Responsible for assessing the contributions of the subnet and ensuring their correctness, thereby earning rewards. Validators can stake TAO tokens to earn additional income.

The emission mechanism of the subnet ( determines the distribution of TAO tokens within the network. Typically, 18% is allocated to subnet owners, 41% to validators, and 41% to miners. The subnet contains 256 UDI slots, with 64 allocated to validators and 192 to miners. Only the 64 validators with the largest staking amounts can be permitted and considered active validators.

There will be a 7-day immunity period after subnet registration. The initial registration fee is 100 TAO, and the price will double for subsequent registrations but will decrease over time. When all subnet positions are filled, newly registered subnets will replace those with the lowest emissions that are not in the immunity period. Therefore, subnets need to continuously increase the validator staking amount and miner efficiency to ensure they are not eliminated.

Thanks to Bittensor's subnet architecture, the decentralized AI data network Masa has successfully landed, becoming the first dual-token reward system in the network, and has secured $18 million in financing.

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

Bittensor's Consensus and Proof Mechanism: Ensuring Network Security and Efficiency

The Bittensor network employs various innovative consensus and proof mechanisms to ensure the security of the network, the quality of data, and the efficient utilization of computing resources.

The Proof of Intelligence mechanism ) PoI ( is Bittensor's original verification and incentive mechanism. In this mechanism:

• Miner nodes prove their work by completing intelligent computing tasks, which may include natural language processing, data analysis, machine learning model training, etc.
• Validators assign tasks to miners and score them based on the quality of completion.

Yuma consensus is the core consensus mechanism of the Bittensor network. In this mechanism:

• Validators give scores based on task completion and input the scores into the Yuma consensus algorithm.
• Validators with a higher number of staked TAO have a higher weight in scoring.
• The algorithm will filter out scores that deviate significantly from the results of the majority of validators.
• The system ultimately allocates token rewards based on the comprehensive score.

An important feature of the Yuma consensus is the "data unawareness principle," which ensures privacy and security in the data processing process. Nodes can perform computations and verifications without needing to understand the specific content of the data being processed.

In addition, Bittensor also introduces the MOE) mixture of experts( mechanism, integrating multiple expert-level sub-models within a single model architecture. Each expert model has relative advantages when addressing specific domain problems. When new data is introduced, different sub-models can work collaboratively to achieve better results than a single model.

With the cooperation of the Yuma consensus mechanism, validators can score and rank expert models, and allocate token rewards, thereby incentivizing continuous optimization and improvement of the models.

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

Bittensor Subnet Ecosystem: A Blooming of AI Applications

As of now, the number of Bittensor subnet registrations has reached 45, of which 40 have been named. In the past, when the number of subnets was limited, registration competition was exceptionally fierce, with registration prices soaring to as high as one million dollars. As Bittensor gradually opens up more subnet slots, the newly registered subnets may not be as stable or effective as the subnets that have been running for a longer time. However, thanks to the subnet elimination mechanism, in the long run, this will be a process of survival of the fittest, making it difficult for poorly performing subnets to survive.

Currently, aside from the root subnet, subnets 19, 18, and 1 are the most关注, with emission ratios of 8.72%, 6.47%, and 4.16% respectively.

The subnet Vision No. 19 focuses on Decentralization image generation and inference, providing top-tier open-source LLM, image generation models, and other resources. Currently, its slot registration fee is 3.7 TAO, and the total earnings of nodes in 24 hours is approximately 627.84 TAO. If the new registered nodes reach the average level, the daily earnings can reach 2.472 TAO, approximately 866 US dollars.

The 18th subnet Cortex.t, developed by Corcel, is dedicated to building a cutting-edge AI platform that provides high-quality text and image responses through APIs. Currently, the slot registration fee is 3.34 TAO, and the total earnings of nodes in 24 hours are approximately 457.2 TAO. New registered nodes can earn up to 1.76 TAO daily if they reach the average level, which is about 553.64 dollars.

Subnet 1 is a text generation decentralized subnet developed by the Opentensor Foundation. As the first subnet project of Bittensor, it has faced considerable skepticism. In March this year, Eric Wall, the founder of Taproot Wizards, even referred to the TAO token as a meme coin in the AI field, pointing out that subnet 1 generates similar results through AI from a large number of nodes when answering questions, which does not improve the actual problem-solving efficiency.

In addition to these generative models, the Bittensor ecosystem also includes various types of data processing large models, trading AI models, and more. For example, subnet 22 Meta Search provides market sentiment analysis by analyzing Twitter data, while subnet 2 Omron utilizes deep neural networks to learn and optimize staking strategies.

From the perspective of收益风险, if the slots can run successfully for more than a few weeks, the收益 is quite considerable. However, new registered nodes will find it difficult to survive in fierce competition if they cannot use high-performance graphics cards and optimize local algorithms.

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

Future Outlook: Opportunities and Challenges Coexist

Looking to the future, Bittensor and the Web3+AI field it represents face enormous opportunities and challenges:

1. Market enthusiasm continues: The heat around AI concepts is on par with Web3, even attracting some funds that might have originally flowed into Web3. The Web3+AI sector is expected to maintain a market focal position for a considerable period.

2. The advantages of the project are obvious: Bittensor is not a traditional VC-led project, and since its launch, it has seen significant growth, while also having both technical and market support.

3. Technological Innovation Leads: Bittensor breaks the situation where Web3+AI projects operate independently. Its unique subnet architecture lowers the barrier for AI teams to migrate to Decentralization networks. The competitive elimination mechanism also forces subnet projects to continuously optimize models and increase staking amounts.

4. Potential risks exist: As the quota for subnets increases and the registration threshold lowers, it may lead to an influx of low-quality projects. At the same time, as the number of subnets increases, the TAO rewards obtained by a single subnet will gradually decrease. If the price of TAO tokens cannot rise accordingly, it may result in lower-than-expected returns.

Overall, Bittensor represents an important attempt at the integration of AI and blockchain technology. It provides an innovative platform for the development of Decentralization AI, with the potential to reshape the landscape of the AI industry. However, to stand out in the fierce market competition and maintain long-term growth, Bittensor needs to continuously innovate, optimize its technology and economic model, while effectively managing potential risks. In the future, as more quality projects join and technology further matures, the Bittensor ecosystem is expected to play an increasingly important role in the Web3+AI field.

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