The Potential of AI Data Tracks and the Rise of Web3 DataFi

In an era where the world is competing to build the best foundational models, computing power and model architecture are indeed important, but the real moat is the training data. This month, the most eye-catching event in the AI circle is undoubtedly Meta demonstrating its strength by assembling a luxury AI team primarily composed of Chinese research talents. The team is led by Alexander Wang, who is only 28 years old and founded Scale AI. Scale AI is currently valued at $29 billion and provides data services to several AI giants, including the U.S. military, OpenAI, Anthropic, and Meta.

The reason Scale AI can stand out among many unicorns is that it recognized the importance of data in the AI industry early on. Computing power, models, and data are the three pillars of AI models. If we compare a large model to a person, then the model is the body, computing power is the food, and data is the knowledge and information.

With the rapid development of LLMs, the industry's focus has gradually shifted from models to computing power. Nowadays, most models have established transformers as their framework. Major players either build their own supercomputing clusters or sign long-term agreements with cloud service providers. After addressing the basic requirements for computing power, the importance of data has become increasingly prominent.

Scale AI is not only dedicated to mining existing data but is also focusing on long-term data generation business. The company is attempting to provide higher quality training data for AI models through teams of human experts from different fields.

Model training is divided into two stages: pre-training and fine-tuning. Pre-training is similar to the process of a baby learning to speak, requiring a large amount of text, code, and other information crawled from the internet. Fine-tuning is akin to school education, with clear goals and directions. Accordingly, the required data is also divided into two categories: one category consists of large amounts of data that do not require much processing, while the other category requires careful design and selection to cultivate specific capabilities of the model.

As the capabilities of models continue to improve, various more refined and specialized training data will become key influencing factors for model performance. In the long run, AI data is also a field that has a snowball effect; with the accumulation of preliminary work, data assets will have the ability to generate compound interest, becoming more valuable over time.

Web3 DataFi: The Ideal Fertile Ground for AI Data

Compared to traditional data companies, Web3 has a natural advantage in the AI data field, giving rise to the new concept of DataFi. The advantages of Web3 DataFi are mainly reflected in the following aspects:

1. Smart contracts ensure data sovereignty, security, and privacy.
2. Distributed architecture attracts the most suitable workforce globally.
3. Clear blockchain incentive and settlement mechanisms
4. Build an efficient, open one-stop data marketplace

For ordinary users, DataFi is the easiest decentralized AI project to participate in. Users do not need expensive hardware investment or a professional technical background; they can participate by simply completing tasks such as providing data, evaluating model outputs, etc.

The Potential Projects of Web3 DataFi

Currently, multiple DataFi projects have secured substantial funding, demonstrating immense potential:

1. Sahara AI: Committed to building a decentralized AI infrastructure and trading marketplace.
2. Yupp: AI model feedback platform where users can evaluate the output quality of different models.
3. Vana: Transforming users' personal data into monetizable digital assets
4. Chainbase: Focused on on-chain data, covering over 200 blockchains.
5. Sapien: Transforming human knowledge into high-quality AI training data
6. Prisma X: Robot Open Coordination Layer, Focus on Physical Data Collection
7. Masa: A leading subnet project in the Bittensor ecosystem that provides real-time data access.
8. Irys: Focused on programmable data storage and computation
9. ORO: Empowering ordinary people to participate in AI contributions
10. Gata: Decentralized data layer that provides various data collection and processing tools.

Although these projects currently have low barriers to entry, the platform advantages will quickly form with the accumulation of users and ecosystem stickiness. In the early stages, emphasis should be placed on incentive measures and user experience to attract enough users. At the same time, project parties need to consider how to manage labor and ensure data quality to avoid the situation where bad money drives out good.

In addition, improving transparency and accelerating the decentralization process are also significant challenges faced by these projects. The large-scale adoption of DataFi requires attracting both individual users and mainstream enterprise clients to form a complete ecological closed loop.

DataFi represents the process of long-term cultivation of machine intelligence by human intelligence, while ensuring the benefits of human labor through smart contracts. For those who are optimistic about the AI era yet hold onto blockchain ideals, participating in DataFi is undoubtedly a wise choice that aligns with the trend.