AI + Web3: Towers and Squares

In the past two years, the development of AI has been accelerated as if a fast-forward button has been pressed. This butterfly effect spurred by Chatgpt has not only opened up a new world of generative artificial intelligence but has also created huge waves in the Web3 field.

With the support of AI concepts, financing in the cryptocurrency market has significantly increased. In the first half of 2024 alone, 64 Web3+AI projects completed financing, with Zyber365 raising $100 million in Series A funding.

The secondary market is more prosperous, with the total market value of the AI track reaching $48.5 billion and a 24-hour trading volume close to $8.6 billion. Significant benefits have arisen from mainstream AI technology advancements, such as the average price of the AI sector rising by 151% after the release of OpenAI's Sora model. The AI effect has also radiated to the Meme sector: the first AI Agent concept MemeCoin GOAT has quickly gained popularity, with a valuation of $1.4 billion.

The research and topics related to AI+Web3 are equally hot, from AI+Depin to AI Memecoin, and then to AI Agent and AI DAO, the speed of the new narrative rotation is dazzling.

The combination of AI + Web3, filled with hot money and futuristic fantasies, inevitably appears to be a marriage arranged by capital matchmaking. It is difficult for us to discern whether beneath this glamorous surface lies the revelry of speculators or the eve of genuine technological innovation.

To answer this question, the key lies in considering whether both parties can benefit from and promote each other. This article will examine how Web3 can play a role in various aspects of the AI technology stack, as well as what new opportunities AI can bring to Web3.

Opportunities in Web3 under the AI Stack

Before discussing this topic, we need to understand the technical stack of AI large models:

1. Data Collection and Preprocessing
2. Model Training and Fine-tuning
3. Inference and Application

In response to the pain points of AI at various stages, Web3 has currently formed a multi-layered, interconnected ecosystem.

Basic Layer: Sharing Economy of Computing Power and Data

Hash Rate

A significant cost of AI lies in the computing power and energy required for training and inference. Taking Meta's LLAMA3 as an example, it requires 16,000 NVIDIA H100 GPUs running for 30 days to complete training, with hardware investment ranging from $400 million to $700 million, and a monthly energy consumption of approximately 1.6 billion kilowatt-hours.

The earliest intersection of Web3 and AI is in the computing power sharing DePin project. Its core logic is: allowing individuals or entities with idle GPU resources to contribute computing power in a decentralized way, improving GPU utilization through a buyer-seller market similar to Uber, and providing users with low-cost and efficient computing power. At the same time, a staking mechanism is used to punish violations.

Main features:

• Gather idle GPU resources: such as idle computing power from third-party data centers, cryptocurrency mining farms, etc.
• Long-tail market oriented towards AI computing power: more suitable for inference rather than training, meeting the needs of small and medium computing power.
• Decentralized Ownership: Resource owners retain control and can flexibly adjust their earnings.

Representative projects include io.net, Aethir, Akash, Render Network, etc.

Data

Data is the cornerstone of AI. Currently, the dilemma of AI data demand is mainly reflected in:

• Data hunger: requires massive data input
• Data quality requirements have increased
• Privacy and Compliance Issues
• High processing costs

Web3 solutions include:

1. Data Collection: Obtain user private data at low cost through incentive mechanisms, such as Grass, Vana, etc.

2. Data preprocessing: Using decentralized incentive mechanisms for data labeling, such as Synesis, Sapien, etc.

3. Data Privacy and Security: Utilizing technologies such as TEE, FHE, and ZK to protect sensitive data, such as Super Protocol, BasedAI, etc.

4. Data Storage: Addressing on-chain data availability issues, such as 0g.AI, etc.

Middleware: Model Training and Inference

Decentralized Marketplace for Open Source Models

Web3 proposes to establish a decentralized open-source model marketplace, tokenizing models and sharing future profits. Projects such as Bittensor, ORA, and Sentient.

Verifiable Reasoning

Using technologies such as ZK proofs to perform permissionless verification of AI model computations on-chain. Key technologies include zkML, opML, TeeML, etc.

Application Layer: AI Agent

The current focus of AI development has shifted from model capability to AI Agents. Web3's contributions in this regard include:

• Decentralization: Establish incentive and punishment mechanisms through PoS and other mechanisms to promote the democratization of the Agent system.
• Cold Start: Helping potential AI Agent projects secure early funding

Such projects as Virtual Protocol, Spectral, etc.

AI Empowering Web3

The impact of AI on Web3 is evident, mainly reflected in:

AI and On-Chain Finance

1. AI and Cryptocurrency Economy: AI Agents can autonomously execute on-chain transactions, assist in asset management, optimize trading experience, etc.

2. AI and On-Chain Transaction Security: Enhancing Transaction Monitoring and Risk Analysis Capabilities

AI and On-chain Infrastructure

1. AI and On-chain Data: Providing Smarter Data Analysis Tools

2. AI and Development & Auditing: Simplifying the Smart Contract Development Process and Improving Audit Efficiency

New Narratives of AI and Web3

1. NFT: Injecting creativity into generative NFTs

2. GameFi: Improve the efficiency and innovation of game content production.

3. DAO: Assisting community management and decision-making

The Significance of the Combination of AI and Web3: Towers and Squares

The relationship between AI and Web3 can be metaphorically described as a tower and a square. AI represents highly centralized technological capabilities, while Web3 represents a decentralized innovative ecosystem.

The significance of the combination of the two is:

• Web3 provides a more transparent and fair development environment for AI through mechanisms such as token economics and decentralized governance.
• AI injects new vitality into Web3, lowers the entry barrier, and brings more innovative possibilities.

Although AI and Web3 have different development paths, their ultimate goal is to better serve humanity through technology. We look forward to seeing more surprising sparks created by the collision of AI and Web3.