IOSG ｜ Expand INFRA new narrative from AI X Web3 technology stack

Foreword

Recently, the rapid development of large language models (LLMS) has triggered interest in using artificial intelligence (AI) to transform various industries.The blockchain industry is not spared,AI X CryptoThe appearance of narrative has attracted much attention.This article discusses the three main methods of integrating AI and encryption, and discusses the unique opportunities of blockchain technology in solving the AI industry.

Three ways of AIXCRYPTO include:

1. Integrate AI into existing products: companies like DUNE are using AI to enhance their products, such as introducing SQL Copilot to help users write complex query.
2. Construct AI infrastructure for the encryption ecosystem: new companies such as Ritual and Autonolas focus on the development of AI -drive infrastructure, and tailor -made for the needs of the encryption ecosystem.
3. Use blockchain to solve the AI industry problems: projects such as Gensyn, EZKL, and IO.NET are exploring how blockchain technology solves the challenges facing the AI industry, such as data privacy, security, and transparency.

The uniqueness of AI X Crypto is that blockchain technology is expected to solve the internal problems in the AI industry.This unique exchange point has opened up new possibilities for innovation solutions, which is beneficial to AI and blockchain communities.

In depth to discuss the field of AI X Crypto, we aims to identify and display the most promising application of blockchain technology in solving the AI industry’s challenges.By cooperating with experts in the AI industry and encrypted builders, we are committed to promoting the development of cutting -edge solutions that make full use of the advantages of two technological advantages.

1. Industry overview

The field of AI x Crypto can be divided into two categories: infrastructure and applicationEssenceAlthough some existing infrastructure continues to provide support for AI use cases, new participants are launching a new AI native architecture in the market.

1. 1 Calculate the network

In the field of AIXCRYPTO, computing networks play a vital role in providing infrastructure required for AI applications.These networks can be divided into two types based on their support tasks: general -purpose computing networks and dedicated computing networks.

1.1.1 General Computing Network

The general -purpose computing network (e.g. IO.NET and Akash) provides users with opportunities to access machines through SSH and provides command line interface (CLI) to build their own applications.These networks are similar to virtual special servers (VPS) and provide personal computing environment in the cloud.

IO.NET is based on the Solana ecosystem and focuses on GPU leasing and computing clusters. Akash based on the COSMOS ecosystem mainly provides CPU cloud server and various application templates.

Views of iOSG Ventures:

Compared with the mature Web2 cloud market, the computing network is still in the early stages.The web3 computing network is not as good as the “Lego” construction module of the Web2, such as the serverless function, VPS and database cloud projects based on major cloud service providers (such as AWS, Azure and Google Cloud).

The advantages of calculating the network include:

Blockchain technology can use unused computing resources and personal computers to make the network more sustainable.
Point-to-point (P2P) design allows individuals to monetize unused computing resources and provide lower cost computing, thereby potentially reduced the cost of 75%-90%.

However, due to the following challenges, it is difficult to invest in actual production and replace Web2 cloud services:

Although pricing is the main advantage of the general -purpose computing network, it is still challenging to compete with mature Web2 cloud companies in terms of function, security and stability.
The point -to -point style may limit the ability of these networks to quickly deliver mature and stable products.Decentralization characteristics will increase development and maintenance costs.

1.1.2 Dedicated computing network

The dedicated computing network adds an additional layer on the basis of the general computing network. Users can deploy specific applications through configuration files.These networks aims to meet specific use cases, such as 3D rendering or AI reasoning and training.

Render is a professional computing network focusing on 3D rendering.In the field of AI, new players like Bittensor, Hyperbolic, Ritual, and Fetch.ai focus on AI reasoning, while FLOCK and Gensyn mainly focuses on AI training.

Views of iOSG Ventures:

The advantages of a dedicated computing network:

Decentralization and Crypto features solve the problem of centralization and transparency in the AI industry.
There is no computational network and verification scheme to ensure the effectiveness of the reasoning and training process.
Privacy protection technology, such as Federal Learning adopted by FLOCK, allow individuals to contribute data for model training, while maintaining their data in local and private.
By supporting smart contracts and downstream blockchain applications, AI reasoning can be used directly on the blockchain.

Source: iosg Ventures

Although the special AI reasoning and training computing network is still in the early stages, we expect that web3 AI applications will give priority to web3 AI infrastructure.This trend has been significantly obvious in the cooperation between Story Protocol and Ritual and MyShell as an intellectual property.

Although the killer -grade applications built by these emerging AI X Web3 infrastructure have not yet appeared, the growth potential is huge.With the maturity of the ecosystem, we expect more innovative applications that use the unique capabilities of decentralized AI computing networks.

2. Data

Data plays a vital role in the AI model. The development of the AI model involves data, including data collection, training data set storage and model storage.

2.1 Data Storage

Decentralization storage AI model is essential to provide inference APIs in a decentralized manner.The reasoning node should be able to retrieve these models from anywhere.As the AI model may reach the size of hundreds of GB, a powerful decentralized storage network is required.Leaders in the field of decentralized storage, such as Filecoin and Arweave, may provide this feature.

Views of iOSG Ventures:

There are huge opportunities in this field.

For the decentralized data storage network optimized by AI models, it provides functions such as version control, storing different low -precision model quantification, and rapid download of large models.
Decentralized vector database, because they are often tied with models, providing more accurate answers by inserting necessary knowledge with problems.The existing SQL database can also add vector search support.

2.2 Data collectionSet and mark

Collecting high -quality data is essential for AI training.Projects based on blockchain, such as Grass, use crowdsses to collect data for AI training and use personal networks.By appropriate incentives and mechanisms, AI trainers can obtain high -quality data at a lower cost.Projects such as TAI-DA and SAIPEN focus on data marks.

Views of iOSG Ventures:

Our observations of this market:

Most data label projects are inspired by Gamefi, attracting users to reduce costs with promise as high -quality marking data with promise to reduce costs.
There are no obvious leaders in this field, and SCALE AI dominates the Web2 data mark market.

2.3 Blockchain Data

When training the AI model specifically for the blockchain, developers need high -quality blockchain data, hoping to use it directly in its training process.Spice AI and Space and Time offers high -quality blockchain data with SDK, enabling developers to easily integrate data into their training data pipelines.

Views of iOSG Ventures:

With the increase of the demand for AI models related to blockchain, the demand for high -quality blockchain data will surge.However, most data analysis tools currently only provide data export data in CSV, which is not ideal for AI training.

In order to promote the development of AI models specifically for blockchain, it is crucial that it provides more blockchain -related machine learning operation and maintenance (MLOP) functions to enhance the experience of developers.These functions should enable developers to seamlessly integrate blockchain data directly into their Python -based AI training pipeline.

3. ZKML

Due to the motivation to use less complicated models to reduce calculation costs, the centralized AI provider is facing trust issues.For example, sometimes users think that ChatGPT performs poorly.Later, this was attributed to Openai’s update to improve model performance.

In addition, the content creator has proposed copyright concerns about AI.It is difficult for these companies to prove that specific data is not included in its training process.

Zero -knowledge machine learning (ZKML) is an innovative method that solves the problem of trust related to centralized artificial intelligence providers.By using zero -knowledge proof, ZKML enables developers to prove the correctness of its artificial intelligence training and inference process without leaking sensitive data or model details.

3.1 training

Developers can perform training tasks in zero -knowledge virtual machines (ZKVM), such as virtual machines provided by Risc Zero.This process generates a proof that the verification training is properly carried out and only authorized data is used.The proof as a developer complies with the evidence of proper training specifications and data use rights.

Views of iOSG Ventures:

ZKML provides unique solutions for the use of authorized data in proof model training, and it is usually difficult to achieve under the characteristics of the black box characteristics of artificial intelligence models.
This technology is still in the early stages.Calculating expenses are huge.The community is actively exploring more cases of ZK training.

3.2 reasoning

Compared with its training corresponding objects, ZKML is much longer for reasoning.Several well -known companies have emerged in this field. They each adopt unique methods to make machine learning reasoning without trust and transparency.

GIZA focuses on building a comprehensive machine learning operation (MLOP) platform and builds a vibrant community around it.Their goal is to provide developers with tools and resources for integrated ZKML to reasoning workflows.

On the other hand, EZKL creates a user -friendly ZKML framework to provide good performance and give priority to the development experience.Their solutions are designed to simplify the process of realizing ZKML reasoning so that more developers can easily use it.

Modulus Labs uses different methods that they have developed their own proof systems.Their main goal is to significantly reduce computing expenses related to ZKML reasoning.By reducing the expenses by 10 times, Modulus Labs tries to make ZKML reasoning more practical and efficient in practical applications.

Views of iOSG Ventures:

ZKML is particularly suitable for Gamefi and Defi scenes, which is essential that no trust is needed.
The computing expenses introduced by ZKML make it difficult for large artificial intelligence models to operate efficiently.
The industry is still looking for a large number of DEFI and Gamefi pioneers using ZKML in its products to show its actual application scenarios.

4. Agent network + other applications

4.1 proxy network

The proxy network is composed of a tool and knowledge that performs specific tasks and knowledge. For example, assist in chain transactions.These agents can cooperate with each other to achieve more complex goals.Several well -known companies are actively developing agents and agency networks similar to chat robots.

Sleepless, Siya, MyShell, CharacterX, and Delysium are important participants in building chat robot agents.Autonolas and Chainml are building a proxy network for more powerful cases.

Views of iOSG Ventures:

Agent is crucial to the application of real world.They can better perform specific tasks than GM artificial intelligence.Blockchain provides several unique opportunities for artificial intelligence agents.

Possessing incentive mechanism: Blockchain provides incentive mechanisms through technologies such as non -homogeneous tokens (NFT).With clear ownership and incentive structure, the creators are motivated to develop more interesting and innovative agents on the chain.
The combined availability of smart contracts: Smart contracts on the blockchain can be combined, operating like Lego bricks.The open API provided by smart contracts enables proxy to perform complex tasks that are difficult to achieve in traditional financial systems.This combination enables proxy to interact with various decentralized applications (DAPPS) and use its functions.
Internal openness: By constructing agents on the blockchain, they inherit the inherent openness and transparency of these networks.This creates major opportunities for the combination of different agents, enabling them to cooperate and combine their own ability to solve more complicated tasks.

4.2 Other applications

In addition to the main categories discussed earlier, several interesting artificial intelligence applications in the Web3 field are receiving attention, although they may not be huge enough to form independent categories.These applications span various fields, showing the diversity and potential of artificial intelligence in the blockchain ecosystem.

Image generation: imgnai
Image prompt monetization: nfprompt
Community training artificial intelligence image generation: BOTTO
Chat robots: Kaito, SuperSight, Galaxy, KNN3, AWESOME QA, QNA3
Finance: Numer AI
Wallet: dawn_wallet
Game: Parallel TCG
Education: HOOKED
Safety: FORTA
DID: WorldCoin
Creator Tool: PLAI LAB

5. Promote AIXCRYPTO to web2 users to achieve large -scale adoption

The reason why AI X Crypto is unique because it can solve the most difficult problem in the field of artificial intelligence.Although there is a gap between the current AIXCRYPTO products and Web2 AI products and lack of attractiveness to Web2 users, AIXCRYPTO still has some unique features, only AIXCRYPTO can be provided.

5.1 Calculating resources with cost -effectiveness:

One of the main advantages of AIXCRYPTO is to provide high cost -effective computing resources.With the increase in demand for LLM, the number of developers in the market increase, and the availability and price of the GPU have become more challenging.The price of GPUs has risen sharply and shortages.

The decentralized computing network such as the DEPIN project can help alleviate this problem by using idle computing power, GPUs of small data centers and personal computing devices.Although the stability of the decentralized calculation power may be less than concentrated cloud services, these networks provide high -cost computing equipment in various regions.This decentralization method minimizes the edge delay to ensure a more distributed and more flexible infrastructure.

By using the power of decentralized computing networks, AIXCRYPTO can provide Web2 users with affordable and easy computing resources.This cost advantage is attractive to attracting Web2 users to adopt AIXCRYPTO solutions, and especially when the demand for AI computing continues to grow.

5.2 Give the creator ownership:

Another important advantage of AI X Crypto is to protect the ownership of the creators.In the current field of artificial intelligence, some agents are easily copied.Through simply writing similar hints, you can easily copy these proxy.In addition, the agent in GPT stores is usually owned by centralized companies, rather than the creator, which limits the creator’s ability to control the work and the ability to effectively achieve profitability.

AI X Crypto uses the mature NFT technology in the field of encryption to solve this problem.By expressing agents as NFT, creators can truly own their works and get actual benefits from them.Every time the user interacts with the agent, the creator can get incentives to ensure a fair return on their efforts.The concept of NFT ownership is not only suitable for agents, but also to protect other important assets in the field of artificial intelligence, such as knowledge bases and tips.

5.3 Protecting privacy and rebuilding trust:

Users and creators have privacy concerns about centralized artificial intelligence companies.Users are worried that their data will be abused to train future models, while creators worry that their works are used but lacks appropriate attribution or compensation.In addition, centralized artificial intelligence companies may sacrifice service quality to reduce infrastructure costs.

These problems are difficult to solve through Web2 technology, while AIXCRYPTO uses advanced web3 solutions.Zero -knowledge training and reasoning can be transparent through the data used and ensure the correct model used by proof.Technology such as trusted execution environment (TEE), federal learning, and completely the same state encryption (FHE) to achieve security and privacy to protect privacy and protect privacy.

Through priority, priority and transparency, AIXCRYPTO enables artificial intelligence companies to re -gain public trust and provide artificial intelligence services that respect user rights, so that they distinguish them from traditional web2 solutions.

5.3 Protecting privacy and rebuilding trust:

5.4 Tracking content source

As the content of artificial intelligence generates becomes more precise, it is more difficult to distinguish the text, images or videos generated by human creation and artificial intelligence.To prevent the abuse of the content of artificial intelligence, people need a reliable way to determine the source of the content.

The blockchain performed well in tracking content sources, just like the success of the supply chain management and NFT.In the supply chain industry, the entire life cycle of blockchain tracking products can identify manufacturers and key milestones.Similarly, the blockchain tracking creators and preventing piracy in the case of NFT are prevented from piracy. Due to its publicity, NFT is particularly vulnerable to piracy.Although there is such vulnerability, the use of blockchain can minimize the losses caused by fake NFT, because users can easily distinguish between true and false tokens.

Through the application of blockchain technology to track the source of artificial intelligence generating content, AIXCRYPTO can provide users with the ability to verify the content of the content of the content of artificial intelligence or human beings, thereby reducing the possibility of abuse and increasing trust in content authenticity.

5.5 Use cryptocurrency development model

Design and training models, especially large models, is an expensive and time -consuming process.The new model is still uncertain, and developers cannot predict their performance.

Cryptocurrency provides a friendly way for developers, which can collect pre -training data, collect reinforcement feedback, and raise funds from interest.This process is similar to the life cycle of typical cryptocurrency projects: funds for private investment or take -off Taiwan, and put to the active contributor to the active contributor at the beginning.

The model can adopt a similar method to raise funds for the sale of tokens for training and the contributor to data and feedback.Through the carefully designed token economic model, this workflow can help individual developers train new models easier than before.

6. The challenge of tokennomics

The AI X Crypto project began to target Web2 developers as potential customers, because encryption has unique value propositions, and the market size of the web2 artificial intelligence industry is considerable.However, for Web2 developers who are unfamiliar with tokens and are unwilling to get involved in token systems, tokens may become an obstacle.

In order to cater to Web2 developers and reduce or remove the practicality of the tokens may cause trouble for Web3 enthusiasts, because this may change the fundamental position of the AI X Crypto project.When trying to integrate valuable tokens into the artificial intelligence SaaS platform, it is a challenging task to find the balance between attracting Web2 developers and maintaining the practicality of the tokens.

In order to facilitate the gap between Web2 and Web3 business models and maintain token value at the same time, the following potential methods can be considered:

Use tokens in the distributed infrastructure network of the project.Implement pledge, rewards and punishment mechanisms to protect the basic network.
Use the token as a payment method, and at the same time provide the use of the use of web2 users
Implement token -based governance
Share income with token holders
Use income to repurchase or destroy tokens
For the services provided by the project, provide discounts and additional functions for token holders

By carefully designing the token economic model that conforms to the interests of Web2 and Web3, the AI X Crypto project can successfully attract Web2 developers while maintaining the value and practicality of its token.

7. Our favorite AI XCryptoScene

Our favorite AI X Crypto scene uses the power of user collaboration to complete tasks in the field of artificial intelligence with the help of blockchain technology.Some specific examples include:

1. Data contribution of AI training, Alignment, and benchmark test (such as Chatbot Arena)

2. Cooperate to build a large -scale sharing knowledge base for various agents (for example, Sahara)

3. Use personal resources to capture network data (for example, GRASS)

By using the collective efforts of blockchain -based incentives and coordination, these models show the potential of decentralization and community -driven methods for AI development and deployment.

in conclusion

We are in the dawn stage of AI and Web3. Compared with other industries, the integration of artificial intelligence and blockchain fields is still early.In the top 50 Gen AI products, there are no products related to Web3.The top LLM tools are related to content creation and editing, mainly for sales, conferences and notes/knowledge bases.Considering a large number of research, documents, sales and community work in the web3 ecosystem, it provides huge potential for the development of custom LLM tools.

At present, developers are focusing on building infrastructure and introducing advanced AI models into chains, although we have not reached the goal.As we continue to develop this infrastructure, we are also exploring the best user scenarios to provide AI reasoning on the chain in a safe and unscrupulous manner, which provides unique opportunities for the blockchain field.Other industries can directly use existing LLM infrastructure for reasoning and fine -tuning.Only the blockchain industry needs its own native AI infrastructure.

In the near future, we expect that blockchain technology will use its advantages to solve the most challenging problems in the artificial intelligence industry, making the AI model more affordable, easy to access and profitable.We also look forward to the narrative of the encryption field with the AI industry, although slightly delayed.In the past year, we have witnessed the combination of developers to combine Crypto, proxy and LLM model.In the next few months, we may see more model models, text and video generation, and 3D generation affect the field of Crypto.

The entire AI and Web3 industries have not received sufficient attention. We urgently look forward to a Cryptoxai killer application in Web3.