Web3+AI: 重塑数字资产化

近年来，随着区块链技术的不断发展，Web3+AI的结合逐渐成为数字经济领域的热门话题。在这个领域，数字资产化被赋予了新的内涵，从基础设施到应用层，各个方面都展现出了前所未有的活力和潜力。

1、算力层：算力资产化

在Web3+AI中，算力被视为一种重要的数字资产。传统上，算力主要由中心化机构掌控，而现在，通过区块链技术，算力可以实现去中心化，并以代币激励的方式扩展网络规模。这种新型的算力资产化为数字经济带来了更多的可能性。

• 区块链上的分布式计算平台（例如Golem、Fantom等）；

• 用于AI模型训练的分布式计算网络（例如SingularityNET、Hyperbolic等）；

• 用于3D渲染的去中心化算力（例如The Render Network等）。

Web3+AI的算力资产化，核心优势在于去中心化算力类项目，结合代币激励很容易扩展网络规模，而且其计算资源成本低，具有高性价比，可满足部分中低端的算力需求。

2、数据层：数据资产化

数据是AI的石油、血液。如果不依赖Web3，一般只有巨头企业手中才有大量的用户数据，普通的创业公司很难获取广泛的数据，用户数据在AI行业的价值也并没有反馈给用户。通过Web3+AI，可以让数据收集、数据标注、数据分布式存储等流程更低成本、更透明、更有利于用户。

收集高质量数据是AI模型训练的前置条件，通过Web3的方式，可以利用分布式网络，结合适当的Token激励机制，采用众包收集的方法，以较低成本获取高质量且广泛的数据。

根据项目用途，数据类项目主要包括以下几类：

• 数据收集类项目（例如Grass等）；

• 数据交易类项目（例如Ocean Protocol等）；

• 数据标注类项目（例如Taida、Alaya等）；

• 区块链数据源类项目（例如Spice AI、Space and time等）；

• 去中心化存储类项目（例如Filecoin、Arweave等）。

数据类Web3+AI项目，在设计Token经济模型的过程中更有挑战性，因为数据比算力更难标准化。

3、平台层：平台价值资产化

平台类项目大多数会对标Hugging Face，以整合AI行业各类资源为核心。建立一个平台，聚合数据、算力、模型、AI开发者、区块链等各种资源和角色的链接，以平台为中心，更便捷地解决各种需求。比如Giza，专注于构建全面的zkML运营平台，旨在使机器学习的推理变得可信和透明，因为数据和模型黑盒是目前AI中普遍存在的问题，通过Web3的方式采用ZK、FHE等密码学技术来验证模型的推理确实有正确执行，是迟早会被行业内呼吁的。

也有做Focus AI的layer1/ layer2，例如Nuroblocks、Janction等。核心叙事是连接了各类算力、数据、模型、AI开发者、节点等资源，通过包装通用组件、通用SDK的方式，帮助Web3+AI类应用实现快速构建和发展。

还有Agent Network类的平台，基于这类平台可以为各种应用场景构建AI Agent，例如Olas、ChainML等。

平台类的Web3+AI项目，主要以Token捕获平台价值的方式，激励平台各参与方共建。对于初创项目从0-1的过程比较有帮助，可以减少项目方寻找算力、数据、AI开发者社区```html

Sovereignty in the new era Huang Renxun put forward a word sovereignty in his speech in Dubai. It is reported that this speech focuses on how sovereignty conforms to the interests and demands of the community, and it is possible to build sovereignty in a new form. In an article, he elaborated in detail the synergistic effect of decentralization with balance, which is opaque and brings transparency, which requires data, while blockchain is conducive to data storage and tracking. Most projects in the whole industrial landscape are using blocks. Chain technology solves the construction problems of industrial infrastructure projects, and a few projects use it to solve some problems of application. The industrial picture is as follows. The production and workflow are as follows. The combination of these links is mainly reflected in four aspects: the computing power level and the capitalization of computing power. In the past two years, the computing power used to train large models has increased exponentially, and it will basically double every quarter at a speed far exceeding Moore's Law. This situation has led to the long-term imbalance between supply and demand of computing power and the rapid rise of hardware prices. However, the cost of computing power is raised, but at the same time, there are a large number of low-end computing hardware idle in the market. Perhaps the single computing power of this part of low-end hardware can not meet the high-performance requirements, but if it is built into a distributed computing network by means of computing power leasing and sharing, it can still meet many application needs, because it can significantly reduce the cost of computing power by using distributed idle computing power. The breakdown of computing power layer includes general decentralized computing power, for example, for training. Decentralized computing power of practice, such as decentralized computing power for reasoning, such as reshaping digital assets press releases, has gradually become a hot topic in the field of digital economy in recent years with the continuous development of blockchain technology. In this field, digital assets have been given new connotations, showing unprecedented vitality and potential in all aspects from infrastructure to application layer. Computing power is regarded as an important digital asset traditionally. Controlled by centralized institutions, but now computing power can be decentralized through blockchain technology and the network scale can be expanded by token incentives. This new type of computing power capitalization has brought more possibilities to the digital economy. The distributed computing platform on the blockchain, such as the distributed computing network for model training, such as the decentralized computing power for rendering, is easy to expand the network scale by combining decentralized computing power projects with token incentives. Moreover, Its computing resources are low in cost and cost-effective, which can meet some low-end computing needs. Data layer data is oil and blood. If it is not dependent, only giant enterprises have a large amount of user data. It is difficult for ordinary startups to obtain a wide range of data, and the value of user data in the industry has not been fed back to users. Through processes such as data collection, data labeling, data distributed storage, etc., it is cheaper, more transparent and more conducive to users to collect high-quality data. It is model training. The way of pre-condition passing can use distributed network combined with appropriate incentive mechanism to obtain high-quality and extensive data at a lower cost. According to the purpose of the project, data projects mainly include the following types of data collection projects, such as data trading projects, such as data labeling projects, such as blockchain data sources, such as decentralized storage projects, such as data projects, which are more challenging in the process of designing economic models because data is more than. Computing power is more difficult to standardize platform layer, platform value and capitalization. Most of the platform projects will be marked with the integration of various resources in the industry as the core, establishing a platform to aggregate data computing models, developers' blockchain and other links of various resources and roles, and solving various needs more conveniently with the platform as the center, such as focusing on building a comprehensive operating platform, aiming at making the reasoning of machine learning credible and transparent, because the black box of data and models is a common problem at present, and adopting cryptography technologies. It will be called for and done by the industry sooner or later to verify that the reasoning of the model is indeed carried out correctly. For example, the core narrative is to connect various resources such as computing data model developer nodes, and help the rapid construction and development of class applications by packaging common components. There are also classes of platforms based on this kind of platform, which can be used to build projects such as platforms for various application scenarios, mainly to encourage all participants to build platforms in a way that captures the value of the platform, which is more helpful for the start-up projects and can reduce the need for the project parties to find the computing data developer community. 比特币今日价格行情网_okx交易所app_永续合约_比特币怎么买卖交易_虚拟币交易所平台

作为一名游戏玩家，我们需要能够快速理解游戏规则，并以最高效的方式完成游戏任务。在DEX中，人工智能已经多年来在套利交易中发挥着重要作用。在预测市场中，AI代理可以通过广泛接受大量数据、知识库和信息，训练其模型的分析预测能力，并将其产品化提供给用户，以帮助用户以模型推理的方式对特定事件进行预测，比如体育赛事、总统大选等。

其次是创建可扩展的去中心化的私人AI。由于许多用户担心AI的黑盒问题，担心系统存在偏见，或者担心某些dApps通过AI技术欺骗用户来牟利。这本质上是因为用户对AI的模型训练和推理过程缺乏审查权限和治理权限。但是如果创建一个像Web3项目一样的Web3 AI，社区对这个AI具有分布式治理权，可能会更容易被接受。

截至目前，在Web3 + AI应用层尚未出现天花板很高的白马项目。

总结

Web3 + AI仍处于早期阶段，业内对这个赛道的发展前景也存在分歧，我们将继续关注这个赛道。我们希望Web3与AI的结合能够创造出比中心化AI更有价值的产品，使AI摆脱“巨头控制”、“垄断”等标签，以更加社区化的方式进行“共治AI”。也许在更加紧密的参与和治理过程中，人类对AI会更加“敬畏”，少一些“恐惧”。

```

注册有任何问题请添加 微信：MVIP619 拉你进入群

文章转载注明 网址：https://netpsp.com/?id=63866