The exploration journey of the integration and innovation of AI and Web3

Author: BadBot, IOBC Capital

Web3, as a decentralized, open, and transparent new internet paradigm, has a natural opportunity for integration with AI. Under traditional centralized architectures, AI computing and data resources are strictly controlled, facing numerous challenges such as computational bottlenecks, privacy breaches, and algorithmic black boxes. Web3, based on distributed technology, can inject new momentum into AI development through shared computing networks, open data markets, and privacy computing. At the same time, AI can empower Web3 in many ways, such as optimizing smart contracts and anti-cheating algorithms, aiding its ecosystem construction. Therefore, exploring the combination of Web3 and AI is crucial for building the next-generation internet infrastructure and unlocking the value of data and computing power.

Data-Driven: The Solid Foundation of AI and Web3

Data is the core driving force behind AI development, just as fuel is to an engine. AI models need to digest a large amount of high-quality data to gain deep understanding and strong reasoning capabilities. Data not only provides the training foundation for machine learning models but also determines the accuracy and reliability of the models.

In the traditional centralized AI data acquisition and utilization model, there are several major issues:

• High costs of data acquisition, making it difficult for small and medium-sized enterprises to bear;

• Data resources are monopolized by tech giants, forming data silos;

• Personal data privacy faces risks of leakage and abuse.

Web3 can address the pain points of traditional models with a new decentralized data paradigm.

• Through Grass, users can sell idle networks to AI companies, decentralizing the capture of network data, which, after cleaning and transformation, provides real, high-quality data for AI model training;

• Public AI adopts a "label to earn" model, incentivizing global workers to participate in data labeling through tokens, gathering global expertise and enhancing data analysis capabilities;

• Blockchain data trading platforms like Ocean Protocol and Streamr provide a public and transparent trading environment for both data supply and demand, incentivizing data innovation and sharing.

Nevertheless, there are still some issues with acquiring real-world data, such as inconsistent data quality, processing difficulties, and insufficient diversity and representativeness. Synthetic data may become the star of the Web3 data track in the future. Based on generative AI technology and simulation, synthetic data can mimic the properties of real data, serving as an effective supplement to real data and improving data utilization efficiency. In fields such as autonomous driving, financial market trading, and game development, synthetic data has already shown its mature application potential.

Privacy Protection: The Role of FHE in Web3

In the data-driven era, privacy protection has become a global focus, with regulations like the EU's General Data Protection Regulation (GDPR) reflecting a strict safeguarding of personal privacy. However, this also brings challenges: some sensitive data cannot be fully utilized due to privacy risks, which undoubtedly limits the potential and reasoning capabilities of AI models.

FHE, or Fully Homomorphic Encryption, allows computations to be performed directly on encrypted data without the need to decrypt it, and the computation results are consistent with those obtained from the same calculations on plaintext data.

FHE provides solid protection for AI privacy computing, enabling GPU computing power to execute model training and reasoning tasks in an environment that does not touch the original data. This brings significant advantages to AI companies, allowing them to securely open API services while protecting trade secrets.

FHEML supports encrypted processing of data and models throughout the entire machine learning cycle, ensuring the security of sensitive information and preventing data leakage risks. In this way, FHEML strengthens data privacy and provides a secure computing framework for AI applications.

FHEML complements ZKML, where ZKML proves the correct execution of machine learning, while FHEML emphasizes computing on encrypted data to maintain data privacy.

Computing Power Revolution: AI Computing in Decentralized Networks

The computational complexity of current AI systems doubles every three months, leading to a surge in computing power demand that far exceeds the supply of existing computing resources. For example, training OpenAI's GPT-3 model requires enormous computing power, equivalent to 355 years of training time on a single device. This shortage of computing power not only limits the advancement of AI technology but also makes advanced AI models inaccessible to most researchers and developers.

At the same time, global GPU utilization is below 40%, coupled with a slowdown in microprocessor performance improvements and chip shortages caused by supply chain and geopolitical factors, exacerbating the computing power supply issue. AI practitioners find themselves in a dilemma: either purchase hardware or rent cloud resources, creating an urgent need for an on-demand, cost-effective computing service model.

IO.net is a decentralized AI computing power network based on Solana, aggregating idle GPU resources globally to provide an economical and accessible computing power market for AI companies. Demand-side users can publish computing tasks on the network, and smart contracts will assign tasks to miner nodes contributing computing power. Miners execute the tasks and submit results, receiving points as rewards after verification. IO.net's solution improves resource utilization efficiency and helps address the computing power bottleneck in fields like AI.

In addition to general decentralized computing power networks, there are platforms focused on AI training, such as Gensyn and Flock.io, as well as dedicated computing power networks for AI reasoning, like Ritual and Fetch.ai.

Decentralized computing power networks provide a fair and transparent computing power market, breaking monopolies, lowering application thresholds, and improving computing power utilization efficiency. In the Web3 ecosystem, decentralized computing power networks will play a key role in attracting more innovative dapps to jointly promote the development and application of AI technology.

DePIN: Web3 Empowers Edge AI

Imagine your phone, smart watch, or even smart devices at home all having the capability to run AI—this is the charm of Edge AI. It allows computation to occur at the source of data generation, achieving low latency and real-time processing while protecting user privacy. Edge AI technology has already been applied in critical areas such as autonomous driving.

In the Web3 domain, we have a more familiar term—DePIN. Web3 emphasizes decentralization and user data sovereignty, and DePIN enhances user privacy protection and reduces the risk of data leakage by processing data locally. The native token economic mechanism of Web3 can incentivize DePIN nodes to provide computing resources, building a sustainable ecosystem.

Currently, DePIN is rapidly developing within the Solana ecosystem, becoming one of the preferred public chain platforms for project deployment. Solana's high TPS, low transaction fees, and technological innovations provide strong support for DePIN projects. Currently, the market value of DePIN projects on Solana exceeds $10 billion, with well-known projects like Render Network and Helium Network making significant progress.

IMO: A New Paradigm for AI Model Release

The concept of IMO was first proposed by the Ora protocol, which tokenizes AI models.

In traditional models, due to the lack of a revenue-sharing mechanism, once an AI model is developed and put on the market, developers often find it difficult to obtain continuous revenue from the subsequent use of the model, especially when the model is integrated into other products and services. The original creators find it hard to track usage, let alone derive revenue from it. Moreover, the performance and effectiveness of AI models often lack transparency, making it difficult for potential investors and users to assess their true value, thus limiting the market recognition and commercial potential of the models.

IMO provides a new funding support and value-sharing method for open-source AI models, allowing investors to purchase IMO tokens and share in the revenue generated by the model subsequently. The Ora Protocol uses two ERC standards, ERC-7641 and ERC-7007, combined with AI oracles (Onchain AI Oracle) and OPML technology to ensure the authenticity of AI models and that token holders can share in the revenue.

The IMO model enhances transparency and trust, encourages open-source collaboration, adapts to trends in the crypto market, and injects momentum into the sustainable development of AI technology. The IMO is still in its early experimental stage, but as market acceptance increases and participation expands, its innovation and potential value are worth looking forward to.

AI Agent: A New Era of Interactive Experience

AI Agents can perceive their environment, think independently, and take corresponding actions to achieve set goals. Supported by large language models, AI Agents can not only understand natural language but also plan decisions and execute complex tasks. They can serve as virtual assistants, learning user preferences through interaction and providing personalized solutions. Even without explicit instructions, AI Agents can autonomously solve problems, improving efficiency and creating new value.

Myshell is an open AI-native application platform that provides a comprehensive and user-friendly toolkit, allowing users to configure robot functions, appearance, voice, and connect to external knowledge bases, aiming to create a fair and open AI content ecosystem. Utilizing generative AI technology, Myshell empowers individuals to become super creators. Myshell has trained specialized large language models to make role-playing more human-like; voice cloning technology can accelerate personalized interactions in AI products, reducing voice synthesis costs by 99%, with voice cloning achievable in just one minute. With Myshell's customized AI Agents, applications can currently be found in video chatting, language learning, image generation, and more.

In the integration of Web3 and AI, the current focus is more on exploring the infrastructure layer, such as how to acquire high-quality data, protect data privacy, host models on-chain, improve the efficient use of decentralized computing power, and verify large language models. As these infrastructures gradually improve, we have reason to believe that the integration of Web3 and AI will give birth to a series of innovative business models and services.