Author: Paul Veradittakit, Partner at Pantera Capital; Translation: Golden Finance xiaozou
As the digital revolution continues to transform our world, I see two technologies in particular as having disruptive potential: artificial intelligence (AI) and cryptocurrency. AI has recently risen to the forefront of technological innovation by mimicking human cognitive functions and its ability to learn from data. AI’s applications, from healthcare to entertainment, are massive and transformative. Backed by the power of blockchain technology, cryptocurrency promises a decentralized financial future that empowers individuals and simplifies processes. AI and cryptocurrency are a powerful combination that signals the advent 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 capabilities in data analysis can sift through large amounts of historical data to predict price fluctuations with greater accuracy.
Additionally, as cryptocurrencies gain more mainstream adoption, security becomes critical. The decentralized nature of these digital currencies makes them more susceptible to fraud and hacking. Machine learning models can be trained to detect unusual transaction patterns, strengthening defenses against potential security breaches.
Scalability and transaction speed are other significant challenges in the crypto world. As more and more people join blockchain networks, fast and seamless transactions become essential. Advanced AI algorithms can optimize network traffic, ensuring efficient data flow and faster transaction times.
Finally, energy consumption associated with crypto mining is a global concern. AI can play a key role in optimizing the mining process, reducing energy footprint, and paving the way for a more sustainable crypto ecosystem.
2. Cross-cutting use cases for encryption and AI
(1) Decentralized AI Market: Encryption Advantages
In the AI space, platforms like Hugging Face have received considerable attention for their role in popularizing pre-trained AI models. However, the emergence of decentralized AI marketplaces within the crypto space has introduced a new dimension of democratization and decentralization, even beyond the open source efforts of traditional platforms like GitHub, including web3 code and smart contracts.
Hugging Face and traditional AI model libraries:
As an outstanding AI model library, Hugging Face has played a key role in popularizing cutting-edge AI models to a wide audience. It provides a central platform where users can access, share, and fine-tune pre-trained models. This greatly lowers the entry barrier for AI practitioners and developers, enabling them to use state-of-the-art models to complete various natural language processing (NLP) tasks.
Crypto-driven decentralized AI marketplace:
However, crypto-driven decentralized AI marketplaces, such as those built on blockchain technology, take democratization and decentralization to a new level.
--True ownership and control: In traditional codebases like Hugging Face, while access is democratized, control and ownership of the underlying infrastructure and data remain 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.
--Incentive Mechanism: Crypto markets contain token-based incentive mechanisms that reward the efforts of contributors, from data providers to model developers. This incentivizes collaboration and innovation while ensuring fair distribution of benefits. In contrast, traditional platforms may lack these direct financial incentives, making crypto platforms more attractive to participants.
--Data privacy and security: Blockchain technology ensures a high level of transparency while maintaining data privacy through techniques such as zero-knowledge proofs. This addresses concerns about data breaches, a key consideration for AI applications. Traditional platforms may not be able to provide the same level of privacy guarantees.
--Interoperability: Cryptocurrency-based markets 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), resulting in a more interconnected ecosystem.
Essentially, while platforms like Hugging Face have made significant progress in democratizing access to AI models, crypto-driven decentralized AI marketplaces build on this foundation 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 more equitable distribution of benefits among participants.
(2) AI-enhanced smart contracts:
Traditional smart contracts are pre-coded with specific conditions. By integrating with artificial intelligence, these contracts can adapt and dynamically respond to external data and conditions, giving rise to 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 weather-related insurance policies to farmers. Traditional insurance contracts rely on predefined conditions and manual claims processing, which 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. An AI model is integrated into the smart contract to continuously analyze weather data. This AI model is trained to identify adverse weather conditions that may affect the insured crops, such as droughts or floods.
Dynamic premium adjustment:
Traditionally, insurance premiums are fixed and claims are processed after an event occurs. In this AI-enhanced smart contract, premiums are dynamically adjusted based on AI's real-time assessment of weather conditions. If AI detects a high risk of severe weather that could damage crops, premiums for affected policies are automatically adjusted upward to reflect the increased risk. Conversely, when AI predicts favorable weather conditions, premiums may be reduced, incentivizing more farmers to purchase insurance.
Automatic Payout:
If the AI model detects severe weather conditions that meet predefined payout criteria (e.g., an ongoing drought), it triggers automatic payouts to affected policyholders. The AI monitors whether payouts are made promptly, reducing the need for manual claims processing and associated delays.
(3) On-chain data analysis: using machine learning
Blockchains have a vast amount of transaction data, which is a gold mine for data scientists and machine learning enthusiasts. Companies like CertiK and TokenMetrics use the power of machine learning tools to extract valuable insights from this data to enhance security, improve investment strategies, and generally optimize blockchain efficiency.
(4) Decentralized GPU Sharing: Empowering AI and Earning Cryptocurrency
Decentralized GPU sharing is a brand new concept that revolutionizes the way computing resources are used in the AI and machine learning communities. Just as Filecoin disrupts data storage by incentivizing users to share their unused storage space, decentralized GPU sharing is based on similar principles.
Why donate your GPU?
GPUs (Graphics Processing Units) are essential for training and fine-tuning complex machine learning and AI models. These processes require a lot of computing power, which is time-consuming and resource-intensive 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 will be rewarded with cryptocurrency. This model allows individuals and organizations to access high-performance GPU resources without investing upfront costs to purchase dedicated hardware or relying on a centralized cloud service. It democratizes the use of GPU power, making it more inclusive and cost-effective for AI and machine learning enthusiasts.
Unit Economics:
--The unit economics in the decentralized GPU sharing network revolve around cryptocurrency rewards. When you lend your GPU to the network, you receive tokens as compensation.
The amount of tokens you earn depends on several factors, including the computing power of your GPU, the duration of your contribution, and the network’s demand for GPU resources.
--Decentralized GPU sharing platforms typically have transparent and predefined reward structures, ensuring fair and predictable participation.
Overall, decentralized GPU sharing not only enables individuals and organizations to access powerful computing resources required for AI tasks, but also allows GPU owners to effectively monetize their hardware. This is a win-win model that can promote collaboration, cost-effectiveness, and accessibility within the AI and machine learning ecosystem.
(5) RLHF Token Model: Connecting AI and Incentives
The Reinforcement Learning Based on Human Feedback (RLHF) token model provides an interesting intersection of AI and token-based incentives. This concept is particularly effective in specialized sectors where traditional financial rewards may not work. A full explanation is provided below:
In the RLHF token model, the main idea revolves around using token rewards to incentivize real human users to provide valuable feedback and training for AI systems. Imagine an AI system that needs to be trained for a specific task, such as content moderation on a social media platform.
Traditionally, training AI models usually 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 gain feedback by rewarding them with tokens for their contributions.
For example, Amazon’s Mechanical Turk (MTurk) is a well-known crowdsourcing platform where people complete small tasks in exchange for rewards. In the RLHF token model, users essentially play the role of “AI trainers”, providing feedback similar to how MTurk task completers perform tasks for rewards. The key difference is that in the RLHF token model, users are rewarded with tokens instead of traditional currencies. These tokens have intrinsic value in the ecosystem, opening up avenues for their utility, such as purchasing AI-generated content or accessing advanced AI services. This token-based incentive system not only simplifies the financial process of AI training, but also fosters a dynamic ecosystem where users actively collaborate to enhance AI systems.
3. Concerns about the integration of AI and encryption
While the combination of AI and cryptocurrency holds great promise, it also presents a host of challenges. AI models, especially deep learning models, are often referred to as “black boxes” due to their opaque nature. Combining this opacity with the transparency of blockchain can raise concerns about accountability and trust. Addressing these issues is critical to establishing a secure and trusted environment at the intersection of these technologies.
Furthermore, ethical considerations become critical. As open-source AI algorithms are deployed on decentralized networks, questions about liability and compensation will inevitably arise when unexpected negative outcomes occur. 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 certain AI applications. Maintaining the decentralized spirit of blockchain while ensuring compliance with regulations such as GDPR is a unique challenge that requires innovative solutions.
4. Future prospects
As artificial intelligence and encryption technologies continue to mature, their combination is expected to give rise to unprecedented applications.
(1) zkML (zero-knowledge machine learning):
One of zkML’s fundamental capabilities is its ability to cryptographically prove the results produced by a specific machine learning model without revealing the model’s details, 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, thereby 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 by letting them verify that their AI-driven compliance models adhere to 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 has the potential to reshape content creation and distribution. This innovation addresses a pressing issue in the digital age, namely that 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 origin and integrity of digital content. This will not only help detect and mitigate malicious deep fakes and fraudulent content, but also protect creators’ intellectual property and provide consumers with a reliable source of information. For example, blockchain can prove the legitimacy of political speeches or video evidence in legal proceedings, thereby mitigating misinformation risks and ensuring accountability. In the creative industry, it can track the creation and ownership of musical works, ensuring that artists receive due credit and compensation. In education, it can verify the authorship of academic works, prevent plagiarism, and maintain academic integrity. In journalism, it can verify news articles, helping readers distinguish between real news and AI-generated content. At its core, using blockchain to verify the authenticity of AI-generated content goes beyond technology — it underpins the credibility, transparency and integrity of digital content in the age of AI, reshaping how information is created, consumed and trusted, while addressing real-world challenges.
(3) Privacy and security assurance:
One of the pressing challenges facing enterprises today is uncertainty about what happens to their proprietary data when it is shared 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 there are non-blockchain solutions such as local models, cloud services like Azure, and legal agreements, blockchain has a unique advantage in providing a transparent and tamper-proof record of data interactions. It can enable enterprises to verify that their data remains unreadable during the computational process, providing a higher degree of confidence in data privacy and security. This assurance is critical for industries that deal with sensitive information, such as healthcare or finance, where data confidentiality is of paramount importance.
Essentially, blockchain’s ability to create an immutable ledger can help businesses maintain control over their data and ensure that data remains private and secure when interacting with AI systems, giving businesses peace of mind and compliance with data protection regulations.
5. Existing projects
A group of visionary projects are already taking the lead in combining encryption 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 such as Bloom, OPT, T0pp, GPT-J, and Stable Diffusion. A key focus of their research is to create accessible alternatives to closed-source AI systems while deepening our understanding of the performance of these models. With a visionary approach, Together aims to build a decentralized supercomputer that seamlessly connects global clouds, mining infrastructure, gaming hardware, and laptops, all coordinated through blockchain technology. Together’s vision encompasses fair access to AI and computing, and is expected to have a transformative impact on the field.
（2）Bit tensor
Bittensor attempts to pioneer the development of decentralized artificial intelligence networks, focusing primarily on enhancing data oracles for smart contracts. This cutting-edge technology leverages blockchain to create a secure, efficient, decentralized framework for capturing and delivering real-world data 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 artificial intelligence.
(3) Akash Network
Akash Network is changing cloud computing through its decentralized marketplace, providing a dynamic alternative to traditional cloud providers. It provides users with cost-effective, elastic, and censorship-resistant access to computing resources. What makes Akash unique is its potential impact on artificial intelligence. By decentralizing 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 widespread computing power. Akash's decentralized approach provides support for AI projects, making it a key player in the development of AI technology.
(4) Gensyn Protocol
The Gensyn protocol establishes a decentralized ecosystem for machine learning computing. The protocol enables AI researchers and practitioners to seamlessly distribute their computing workloads. When solvers contribute their computing resources, allowing access to high-performance hardware for training and reasoning, verifiers ensure the accuracy and integrity of AI tasks, which is critical to maintaining the credibility of AI models. In addition, Gensyn prioritizes privacy and security by providing methods such as secure mapping layers and encrypted 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 retain full control over their assets, free from the risks associated with centralized exchanges. One of the platform’s innovative products is a cutting-edge NFT marketplace, powered by AI to enhance pricing, discovery, and authentication, distributing rewards to investors. Additionally, the GPT-3-powered Telegram AI Bot provides real-time financial advice and market analysis to all Telegram users for free. Nexus AI further supports informed investment decisions with Asset Intelligence reports, providing comprehensive insights into crypto assets. Through AI-driven solutions such as the staking dashboard and smart investment bots, Nexus AI enables users to confidently operate in the market and achieve their financial goals with precision and efficiency.
（6）Modulus Labs
Modulus Labs has launched an innovative solution aimed at integrating artificial intelligence into blockchain technology. They have launched a zero-knowledge (ZK) prover tailored for AI applications, solving the challenges 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 to launch significant AI applications on the Ethereum platform through collaboration with partners such as Upshot and Ion Protocol. In addition, they are open sourcing the ZKML prover to the broader community. Modulus envisions a future where cryptographic accountability can ensure the integration of AI into various fields (including finance and healthcare) without compromising security.
(7) Render Network
Render Network is a platform that provides decentralized GPU rendering capabilities. This innovative network enables users to scale GPU rendering work with high-performance GPU nodes worldwide through a blockchain-based market for idle GPU computing resources. The platform's vision is to democratize GPU cloud rendering, making it more efficient, more scalable, and available for next-generation 3D content creation. It addresses challenges related to network scalability, efficient infrastructure utilization, GPU power optimization, and digital rights management for immersive media and artificial intelligence applications. Render Network transforms GPU power consumption into an interconnected decentralized economy of 3D assets, providing solutions for developers in various industries, including media, augmented reality, virtual reality, games, and more.
（8）Ritual
Ritual is a company that stands at the intersection of cryptography and AI, revolutionizing the way developers create, distribute, and enhance AI models. By leveraging 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 variety of AI models through a unified API while benefiting from an encrypted infrastructure that ensures computational integrity and privacy. The platform's first product is Infernet, marking the beginning of a full suite of protocols and utilities that seamlessly integrate with Ritual for developers and applications to gain permissionless access to their models and a network of computing providers. Ritual's vision is to become the AI hub in the web3 world, driving innovation, accessibility, and democratization of the world of AI.
6 Conclusion
The collaboration of AI and crypto is not only a fusion of technologies, but also a fusion of possibilities. It represents a future in which machines think, learn, and transact in a decentralized, secure environment. The road ahead is full of challenges, but every challenge is an opportunity for innovation and progress. Innovators and thinkers standing at this crossroads have the potential to write a chapter that redefines the digital age. As we move forward, we must be careful, wise, and visionary to ensure that we reap the benefits while tackling the challenges, ultimately shaping a brighter, better, and more technologically advanced future for all.
