Pantera Partner: Crypto and AI combined use cases and existing projects

作者：Paul Veradittakit，Pantera Capital合伙人；翻译：金色财经xiaozou

As the digital revolution continues to change our world, I believe there are two technologies in particular that have disruptive potential: artificial intelligence (AI) and cryptocurrency. Artificial intelligence (AI) has recently risen to the forefront of technological innovation by mimicking human cognitive functions and the ability to learn from data. The scale of AI applications, from healthcare to entertainment, is enormous and transformative. Powered by powerful blockchain technology, cryptocurrencies promise a decentralized financial future that empowers individuals and streamlines processes. The combination of artificial intelligence and cryptocurrencies is showing great power, signaling the dawn of a new era of technological symbiosis.

1. Encryption problems that AI can solve

While cryptocurrencies have become a revolutionary means of trading and investing, they are not without their challenges. Market volatility is a pressing concern for investors. AI's ability in data analytics can sift through large amounts of historical data to predict price fluctuations with greater accuracy.

In addition, as cryptocurrencies gain more mainstream adoption, security becomes paramount. The decentralized nature of these digital currencies makes them more susceptible to fraud and hacking. Machine learning models can be trained to detect anomalous transaction patterns and strengthen defenses against potential security breaches.

Scalability and transaction speed are other significant challenges in the crypto world. As more and more people join the blockchain network, fast and seamless transactions become essential. Advanced AI algorithms optimize network traffic, ensuring efficient data flow and faster transaction times.

Finally, the energy consumption associated with crypto mining is a global concern. AI can play a key role in optimizing the mining process, reducing the energy footprint, and paving the way for a more sustainable crypto ecosystem.

2, Cross-use cases for encryption and AI

(1) Decentralized AI Market: Crypto Advantages

In the field of artificial intelligence, platforms like Hugging Face have garnered considerable attention for their role in popularizing pre-trained AI models. However, the emergence of a fragmented AI market within the crypto space has introduced a new dimension of democratization and decentralization, even beyond the open-source efforts made by traditional platforms such as GitHub, including web3 code and smart contracts.

• Hugging Face & Traditional AI Model Library:

Hugging Face, as an outstanding library of AI models, has played a key role in democratizing cutting-edge AI models to a wide audience. It provides a central platform where users can access, share, and fine-tune pretrained models. This dramatically lowers the barrier to entry for AI practitioners and developers, enabling them to leverage state-of-the-art models for a wide range of natural language processing (NLP) tasks.

• Crypto-Powered Decentralized AI Marketplace:

However, crypto-powered decentralized AI markets, such as those built on blockchain technology, take democratization and decentralization to the next level. The details are as follows:

--True ownership and control: In traditional codebases like Hugging Face, while access is democratized, control and ownership of the underlying infrastructure and data remains centralized. In contrast, crypto-based marketplaces typically employ decentralized blockchain networks, ensuring that control is distributed among network participants. Users have a say in management decisions, making it a more democratic and community-driven ecosystem.

--Incentives: The crypto market contains token-based incentives that reward the efforts of contributors, from data providers to model developers. This stimulates collaboration and innovation, while ensuring a fair distribution of benefits. In contrast, traditional platforms may lack these direct financial incentives, which makes crypto platforms more attractive to participants.

--Data privacy and security: Blockchain technology ensures a high level of transparency while maintaining data privacy through technologies such as zero-knowledge proofs. This addresses concerns about data breaches, which is a key consideration for AI applications. Traditional platforms may not offer the same level of privacy guarantees.

--Interoperability: Cryptocurrency-based marketplaces are often created based on blockchain standards and are designed with interoperability in mind. This means that AI models and services can be seamlessly integrated with a variety of blockchain-based applications, smart contracts, and decentralized applications (dApps) to form a more interconnected ecosystem.

Essentially, while platforms like Hugging Face have made significant strides in democratizing AI models, crypto-powered decentralized AI marketplaces are built on top of this by combining blockchain's decentralization, true ownership, and token-based incentives. This increased democratization and decentralization has the potential to reshape the AI landscape by fostering collaboration, innovation, and a more equitable distribution of benefits among participants.

(2) AI-enhanced smart contracts:

Traditional smart contracts are pre-coded with specific conditions. Through integration with artificial intelligence, these contracts can be adaptive, dynamically responding to external data and conditions, enabling more efficient and versatile decentralized applications.

Let's look at a more practical example.

Imagine a decentralized insurance platform based on blockchain technology that provides farmers with weather-related insurance policies. Traditional insurance contracts rely on predefined conditions and manual claims processing, a process that can be slow and prone to disputes. In this case, AI-enhanced smart contracts will come into play and revolutionize the insurance industry.

• Weather data and AI analysis:

Smart contracts are designed to interact with external data sources, such as weather APIs, to obtain real-time weather data for the insured area. The AI model is integrated into the smart contract to continuously analyze the weather data. This AI model is trained to identify adverse weather conditions, such as drought or flooding, that may affect insured crops.

• Dynamic Premium Adjustment:

Traditionally, insurance premiums are fixed and claims are processed after the event. In this AI-enhanced smart contract, premiums are dynamically adjusted based on AI's real-time assessment of weather conditions. If the AI detects a high risk of severe weather that could damage crops, the premiums of the affected policies will be automatically adjusted upwards to reflect the increased risk. Conversely, when AI predicts favorable weather conditions, premiums may be lowered, incentivizing more farmers to buy insurance.

• Automatic Payout:

If the AI model detects severe weather conditions (e.g., persistent drought) that meet predefined payout criteria, it triggers an automatic payout to the affected policyholder. AI monitors whether claims are being made quickly, reducing the need for manual claims processing and associated delays.

(3) On-chain data analysis: Leveraging machine learning

Blockchain, with its vast amount of transactional data, is a goldmine for data scientists and machine learning enthusiasts. Companies like CertiK and TokenMetrics are harnessing the power of machine learning tools to derive valuable insights from this data to enhance security, improve investment strategies, and optimize blockchain efficiency across the board.

(4) Decentralized GPU sharing: Empower AI and earn cryptocurrency

Decentralized GPU sharing is a completely new concept that has revolutionized the way computing resources are used in the AI and machine learning community. In the same way that Filecoin subverts data storage by incentivizing users to share their unused storage space, decentralized GPU sharing is based on a similar principle.

• Why Contribute Your GPU?

GPUs (graphics processing units) are essential for training and fine-tuning complex machine learning and artificial intelligence models. These processes require a lot of computing power, which is a time-consuming and resource-intensive process on a personal computer. By contributing your GPU to a decentralized GPU sharing network, you can contribute your computing resources to a larger pool that anyone can access. In return, you'll be rewarded with cryptocurrency. This model allows individuals and organizations to access high-performance GPU resources without having to invest upfront in dedicated hardware or rely on central cloud services. It democratizes the use of GPU features, making it more inclusive and cost-effective for AI and machine learning enthusiasts.

• Unit Economics:

--Unit economics in a decentralized GPU sharing network revolves around cryptocurrency rewards. When you lend your GPU to the network, you get tokens as compensation.

--The number of tokens you earn depends on several factors, including the computing power of the GPU, the duration of the contribution, and the network's demand for GPU resources.

--Decentralized GPU sharing platforms often have transparent and predefined reward structures to ensure fair and predictable participation.

Overall, decentralized GPU sharing not only gives individuals and organizations access to the powerful computing resources needed for AI tasks, but also allows GPU owners to effectively monetize their hardware. It's a win-win model that fosters collaboration, cost-effectiveness, and accessibility within AI and ML ecosystems.

(5) RLHF Token Model: Connecting AI and Incentives

The Human Feedback-Based Reinforcement Learning (RLHF) token model offers an interesting intersection of artificial intelligence and token-based incentives. This concept is particularly effective in specialized sectors where traditional financial incentives may not work. A full explanation is provided below:

In the RLHF token model, the main idea revolves around the use of token rewards to incentivize human users to provide valuable feedback and training for the AI system. Imagine an AI system that needs to be trained on a specific task, such as content moderation on a social media platform.

Traditionally, training AI models often requires hiring human trainers to label data or fine-tune algorithms, a process that can be very resource-intensive. However, the RLHF token model introduces a more innovative approach. They encourage users to actively participate in the AI system and get feedback by rewarding them with tokens for their contributions.

Amazon's Mechanical Turk (MTurk), for example, is a well-known crowdsourcing platform where people get paid to complete small tasks. In the RLHF token model, users essentially act as "AI trainers", providing feedback similar to MTurk task completionists performing tasks to get paid. The key difference is that in the RLHF token model, users receive tokens as a reward instead of traditional currency. These tokens have intrinsic value in the ecosystem, opening the way for their utility, such as purchasing AI-generated content or accessing advanced AI services. This token-based incentive system not only streamlines the financial process of AI training, but also fosters a dynamic ecosystem within which users actively collaborate to enhance AI systems.

3. Concerns about the convergence of AI and crypto

While the combination of AI and cryptocurrencies holds great promise, it also brings with it its own set of challenges. AI models, especially deep learning models, are often referred to as "black boxes" due to their opacity. Combining this opacity with the transparency of blockchain can raise concerns about accountability and trust. Addressing these issues is critical to building a secure and trusted environment at the intersection of these technologies.

In addition, ethical considerations become crucial. With the deployment of open-source AI algorithms on decentralized networks, liability and compensation-related issues are bound to arise in the event of unintended negative outcomes. Striking the right balance between innovation and ethical responsibility is an ongoing challenge that requires careful consideration.

Another challenge is data privacy. Balancing the transparency and data privacy needs of blockchain can be tricky in some AI applications. Maintaining the decentralized spirit of blockchain while ensuring compliance with rules such as GDPR is a unique challenge that requires innovative solutions.

4. Future prospects

As AI and cryptography continue to mature, the convergence of the two is expected to lead to unprecedented applications.

(1)zkML (Zero-Knowledge Machine Learning):

One of the fundamental features of zkML is its ability to cryptographically prove the results produced by a particular machine learning model without revealing the details of the model, opening up new possibilities for secure and transparent data processing. zkML opens the door to innovative applications for large enterprises, including:

• Financial services: Financial institutions can use zkML to verify the integrity of credit scoring models and loan eligibility decisions without revealing sensitive customer data, enhancing trust and accuracy in the lending process.
• Healthcare: Hospitals and research institutions can use zkML to evaluate the effectiveness of medical diagnostic models while keeping patient data confidential, ensuring data privacy and medical accuracy.
• Legal compliance: zkML can help companies comply with regulations, allowing companies to validate their AI-driven compliance models to comply with industry regulations without exposing proprietary algorithms.
• Smart contracts: In blockchain-based applications, zkML can ensure that smart contracts accurately execute machine learning models, providing trust and transparency for decentralized financial services, insurance, and more.

(2) Authenticity of AI-generated content:

Verifying the authenticity of AI-generated content through blockchain technology is expected to reshape the creation and distribution of content. This innovation addresses a pressing issue in the digital age, where AI can convincingly mimic human-generated content, leading to concerns about misinformation, intellectual property, and trust in digital media. By anchoring content authenticity on the blockchain – a tamper-proof and transparent distributed ledger – consumers, creators, and institutions can easily determine the provenance and integrity of digital content. Not only does this help detect and mitigate malicious deepfakes and fraudulent content, but it also protects creators' intellectual property and provides consumers with a reliable source of information. For example, blockchain can prove the legitimacy of political speeches or video evidence in legal proceedings, mitigating the risk of misinformation and ensuring accountability. In the creative industries, it can track the creation and ownership of musical works, ensuring that artists are given the trust and compensation they deserve. In the field of education, it can verify the authorship of academic works, prevent plagiarism, and maintain academic integrity. In the field of journalism, it can verify news articles and help readers distinguish between real news and AI-generated content. Essentially, the use of blockchain to verify the authenticity of AI-generated content goes beyond technology – it underpins the trustworthiness, transparency, and integrity of digital content in the age of AI, reshaping the way information is created, consumed, and trusted, while addressing real-world challenges.

(3) Privacy & Security Guarantee:

One of the pressing challenges facing businesses today is the uncertainty about how to process data when sharing their proprietary data with AI platforms like OpenAI. This is where data privacy issues arise, including whether the data is used for training, who has access to the data, and whether the data is secure throughout the computational process. While on-premise models, cloud services such as Azure, and non-blockchain solutions such as legal agreements exist, blockchain has unique advantages in providing a transparent and tamper-proof record of data interactions. It allows businesses to verify that their data remains unreadable during computation, providing a higher degree of confidence in data privacy and security. This assurance is essential for industries that deal with sensitive information, such as healthcare or finance, where data confidentiality is critical.

Essentially, blockchain's ability to create an immutable ledger can help businesses maintain control over their data and ensure that data privacy and security are maintained during interactions with AI systems, giving businesses peace of mind and compliance with data protection rules.

5. Existing projects

A group of visionary projects have pioneered the combination of crypto and artificial intelligence.

（1）Together.ai

Together's mission is to revolutionize the world of AI by building an open-source, decentralized cloud platform that caters to large fund models. They have built a sizable decentralized cloud platform dedicated to AI research, with many open models on it, such as Bloom, OPT, T0pp, GPT-J, and Stable Diffusion. A key focus of their research work is to create accessible alternatives to closed-source AI systems, while deepening our understanding of the performance of these models. Through a visionary approach, Together aims to build a decentralized supercomputer that seamlessly connects the global cloud, mining infrastructure, gaming hardware, and laptops, all coordinated through blockchain technology. Together's vision encompasses equitable access to AI and computing that promises to have a transformative impact on the field.

(2)Bittensor

Bittensor seeks to pioneer the development of decentralized AI networks, with a primary focus on augmenting smart contract data oracles. This cutting-edge technology leverages blockchain to create a secure, efficient, decentralized framework for taking real-world data and delivering it to smart contracts. Bittensor's unique decentralized oracle approach significantly improves the accuracy, speed, and reliability of data feeds, making it a core player at the intersection of crypto and AI.

(3)Akash网络

The Akash Network is transforming cloud computing through its decentralized marketplace, offering a dynamic alternative to traditional cloud providers. It provides users with cost-effective, resilient, and censorship-resistant access to computing resources. What makes Akash unique is its potential impact on artificial intelligence. Through decentralized cloud computing, Akash opens the door for AI developers to access scalable, secure, and efficient resources. This is especially important because AI relies heavily on a wide range of computing power. Akash's decentralized approach supports AI projects, making them key players in the development of AI technology.

(4)Reunion协议

The Gensyn protocol establishes a decentralized ecosystem for machine learning computing. The protocol enables AI researchers and practitioners to seamlessly distribute their computing workloads. While the solver contributes its computing resources, allowing access to high-performance hardware for training and inference, the veryfier ensures the accuracy and completeness of the AI task, which is essential for maintaining the credibility of the AI model. In addition, Gensyn prioritizes privacy security by providing methods such as a secure mapping layer and encrypting training data to protect sensitive information in AI applications. The protocol's decentralized nature, efficient scalability, and global accessibility democratize AI, making it more accessible and cost-effective for users.

（5）Nexus AI

Nexus AI is at the forefront of financial empowerment, leveraging advanced AI algorithms to provide investors with unparalleled insights into market trends. What sets Nexus AI apart is its unwavering commitment to decentralization, ensuring that investors maintain full control over their assets from the risks associated with centralized exchanges. One of the platform's innovative offerings is a cutting-edge NFT marketplace powered by artificial intelligence to enhance pricing, discovery, and authentication, distributing rewards to investors. In addition, the Telegram AI Bot, powered by GPT-3, provides real-time financial advice and market analysis for free to all Telegram users. Nexus AI further supports informed investment decisions through Asset Intelligence reports, providing comprehensive insights into crypto assets. With AI-powered solutions such as staking dashboards and smart investment bots, Nexus AI empowers users to operate the market with confidence and achieve their financial goals with precision and efficiency.

(6)Modulus Labs

Modulus Labs has launched an innovative solution that aims to integrate artificial intelligence into blockchain technology. They have launched a zero-knowledge (ZK) prover tailored for AI applications, solving the challenge of ensuring tamper-proof AI access for decentralized applications (dApps). This solution enables smart contracts to leverage powerful AI capabilities without compromising the fundamental principles of decentralization. Modulus is actively working on launching important AI applications on the Ethereum platform through collaboration with partners such as Upshot and Ion Protocol. In addition, they are open-sourced ZKML prover to the wider community. Modulus envisions a future where cryptographic accountability can ensure the integration of AI into all sectors, including finance and healthcare, without compromising security.

(7)Render网络

The Render Network is a platform that provides decentralized GPU rendering capabilities. This innovative network enables users to scale GPU rendering jobs globally with high-performance GPU nodes through a blockchain-based marketplace of idle GPU computing resources. The platform's vision is to democratize GPU cloud rendering, making it more efficient, scalable, and available for next-generation 3D content creation. It addresses the challenges associated with network scalability, efficient infrastructure utilization, GPU power optimization, and digital rights management for immersive media and AI applications. The Render Network transforms GPU power consumption into a decentralized economy of interconnected 3D assets, providing solutions for developers across industries, including media, augmented reality, virtual reality, gaming, and more.

(8)Ritual

Ritual is a company that stands at the intersection of cryptography and artificial intelligence and has revolutionized the way developers create, distribute, and augment AI models. By harnessing the power of encryption, Ritual provides an open, modular, sovereign execution layer for AI. It connects a distributed network of nodes with computing resources and model creators, allowing creators to host their AI models. In turn, users can easily access a wide range of AI models through a unified API, while benefiting from a cryptographic infrastructure that ensures computational integrity and privacy. The platform's first product, Infernet, marks the beginning of a full suite of protocols and utilities for developers and applications to seamlessly integrate with Ritual and gain permissionless access to their models and network of compute providers. Ritual's vision is to be the AI hub in the web3 world, driving innovation, accessibility, and democratization in the AI world.

6. Conclusion

The collaboration of AI and encryption is not only a convergence of technologies, but also a convergence of possibilities. It represents a future where machines think, learn, and trade in a decentralized, secure environment. The road ahead is full of challenges, but every challenge is an opportunity to give birth to innovation and progress. Innovators and thinkers at this crossroads have the potential to write a chapter that redefines the digital age. As we move forward, we must be cautious, wise and far-sighted to ensure that we reap the benefits while tackling challenges smartly, ultimately shaping a brighter, more technologically advanced future for all.