Understanding the Intersection of Crypto and AI: Insights from Galaxy Digital's Lucas Tcheyan

In a rapidly evolving technological landscape, the convergence of cryptocurrency and artificial intelligence has become a focal point of innovation and speculation. On this episode of Unlayered, hosts Sal and Dave engage in an enlightening conversation with Lucas Tcheyan, a researcher at Galaxy Digital, to explore the intricate relationship between these two groundbreaking fields. Tcheyan, known for his comprehensive paper on the intersection of crypto and AI, shares his insights on the current state of the industry, potential future developments, and the challenges that lie ahead.

The Three Layers of Crypto-AI Integration

Tcheyan's research has identified three distinct layers where cryptocurrency and AI technologies are intersecting:

1. Hardware Layer
2. Connective Tissue Layer
3. AI Agents Layer

Each of these layers plays a crucial role in the development of decentralized AI systems and applications.

The Hardware Layer: Decentralized GPU Farms

At the foundation of the crypto-AI integration lies the hardware layer, primarily focused on providing the necessary computational resources for AI processes. Tcheyan explains:

"A lot of people, any of you follow crypto, you've heard of these applications and protocols. What they're really doing is they're just providing the base hardware, whether it's memory, storage, GPUs, so that these AI processes and workflows can actually run on-chain."

This layer is crucial due to the current shortages in chip supply and regulatory concerns surrounding centralized AI development. Projects in this space aim to create decentralized marketplaces for GPU resources, allowing individuals to rent out their unused computing power to those in need of AI processing capabilities.

Some notable projects in this category include:

• Akash Network
• Render Network
• Livepeer
• Golem Network

These platforms enable a more distributed approach to AI computation, potentially addressing issues of centralization and resource scarcity that plague the current AI landscape.

The Connective Tissue Layer: Bridging Blockchain and AI

The second layer Tcheyan identifies is what he calls the "connective tissue" layer. This critical component acts as a bridge between the compute-constrained nature of blockchains and the resource-intensive requirements of AI processes. Tcheyan elaborates:

"Basically what they're doing is they're reconciling those two differences. So they're allowing you to run the compute off chain. So for example, run an inference or chain a model using GPUs and then post back on to chain a proof verifying that the person who said they trained that model did it the way they said they would or with the model they said they would or with the data that they said they would."

This layer is essential for ensuring transparency and verifiability in AI processes while leveraging the security and immutability of blockchain technology. Key developments in this area include:

1. Zero-Knowledge Machine Learning (ZKML): Utilizing zero-knowledge proofs to verify AI computations without revealing sensitive data.
2. Co-processors: Providing streamlined workflows for developers to integrate AI capabilities into blockchain applications.

Notable projects in the connective tissue layer include:

• Modulus Labs
• Ritual
• Allora

These projects are working to create seamless interfaces between blockchain networks and AI systems, enabling developers to build more sophisticated and trustworthy AI-powered applications on decentralized platforms.

The AI Agents Layer: Autonomous Decision-Making in Crypto

The third and perhaps most intriguing layer of the crypto-AI intersection is the development of AI agents. These autonomous entities are designed to interact with blockchain systems and execute complex tasks on behalf of users. Tcheyan describes them as:

"It's literally just an AI model that has been trained on some set of data, whether it's yourself or maybe they train on DeFi data. So they're really good at generating yield farming strategies or something like that. And that you as an individual can interact with to carry out and execute tasks for yourself."

The potential of AI agents in the cryptocurrency space is vast, ranging from automated trading strategies to complex governance decisions in decentralized autonomous organizations (DAOs). One of the most notable projects in this space is Morpheus, which has gained support from industry veterans like Erik Voorhees.

However, Tcheyan acknowledges that the development of truly useful and competitive AI agents in the crypto space is still in its early stages. He notes:

"My general view is we're still building out or still even just experimenting with the foundational primitives that others can iterate and build more complex AI applications and infrastructure on."

The Challenge of Competing with Centralized AI Giants

One of the primary challenges faced by crypto-AI projects is the need to compete with established centralized AI companies like OpenAI and Microsoft. These tech giants have significant resources and data advantages, making it difficult for decentralized alternatives to gain traction.

Tcheyan draws parallels to the early days of decentralized finance (DeFi), suggesting that the true potential of crypto-AI integration may only be realized as the ecosystem matures:

"Similar to what we saw with DeFi, early on in DeFi, it was pretty basic primitives that were cool and interesting but could be easily replicated in traditional finance and it wasn't going to immediately clear like why use these. When I started to get excited about DeFi was when we started to see the DeFi Legos play out and we started to see these protocols iterate on each other and then also figure out ways that they could interact and be composable together to create unique financial products."

He believes that a similar evolution could occur in the crypto-AI space, with projects building on each other's innovations to create novel and compelling use cases that centralized alternatives cannot match.

The Importance of Data in Crypto-AI Development

A critical factor in the success of AI systems is access to high-quality, diverse datasets. In the context of crypto-AI integration, projects are exploring innovative ways to incentivize data sharing and create more robust AI models. Tcheyan highlights the potential of blockchain-based data marketplaces:

"I think we're seeing tokens. We're seeing projects start to like try to implement like BitTensor like models with data you know create very complex like incentive farming schemes to try and get people to provide their data."

He also mentions projects like The Graph, which are working to make blockchain data more accessible and usable for AI training:

"The Graph protocol is doing where they're basically saying like hey we have like indexed most of the blockchain at this point and what we have not done it is indexed it in a way where it's easily readable by machine learning programs."

These efforts to create decentralized data ecosystems could potentially level the playing field between crypto-AI projects and centralized tech giants, enabling the development of more competitive and diverse AI models.

The Evolution of User Interfaces in Crypto-AI Applications

As AI becomes more integrated into cryptocurrency applications, the way users interact with these systems is likely to evolve. Tcheyan discusses the potential for chat-based interfaces and natural language processing to simplify complex crypto operations:

"I think like because the back end of crypto is just so complex for like the normal user and because there's such a focus on like account abstraction servicing intense and everything like that that fits perfectly with what these chat bots and LLMs do because if you can just type into an LLM like hey I want to swap five soul for 10 ETH it's like that's kind of like you're surfacing an intent right."

However, he also acknowledges that the future of user interfaces in crypto-AI applications may not necessarily be text-based. Instead, he suggests that a combination of intuitive design and context-aware AI could lead to more seamless and user-friendly experiences:

"I think the intent based up is interesting where you can customize the actual UX and like the design based on what you think the user wants to do it's like I'll change like the layout the buttons and whatever the interface is that feels like probably some form of that is a future."

This evolution in user interfaces could play a crucial role in making crypto and AI technologies more accessible to a broader audience, potentially driving wider adoption and innovation in the space.

The Role of Solana in the Crypto-AI Ecosystem

As a researcher with a focus on Solana, Tcheyan provides insights into the blockchain's position in the evolving crypto-AI landscape. He highlights Solana's transition to a multi-client architecture as a significant development:

"I think Solana is in an interesting place I think like for me one of the more interesting things is watching it evolve from a single client to a multi-client because I think like one of the most exciting developments on the road on the roadmap for Solana is rolling out fire dancer."

This transition, while potentially slowing down some developments in the short term, is seen as a positive step towards a more robust and decentralized ecosystem. Tcheyan also notes the ongoing technical discussions and innovations happening within the Solana community:

"I'm definitely excited by some of like the deeper tech conversations happening on Solana things like figuring out asynchronous execution, figuring out concurrent leaders, trying to figure out how do we do economic back pressure."

These developments could position Solana as a strong contender in the race to become the preferred blockchain for AI-powered applications and services.

The Changing Landscape of NFTs on Solana

While Solana gained significant attention for its NFT ecosystem, recent trends have seen a shift towards more liquid, token-based trading. Tcheyan reflects on this transition:

"I think once people realize that instead of trading a highly illiquid picture they can trade a much more liquid token that is you know maybe that also has a community element to it. You know in some ways it's just an NFT with a much bigger number of NFTs minted. It's just a better way to operate and a better way to do things on chain."

However, he emphasizes that NFTs are far from dead on Solana, pointing to innovative projects and potential new use cases:

"I don't think NFTs are dead on Solana in any way. I continue to think that like Tensor is one of the most amazing products in crypto. I think what's happening with compressed NFTs open the doors for a lot of things."

Tcheyan suggests that the future of NFTs on Solana may involve a reimagining of their primary use cases, moving beyond speculative PFP collections to more utility-focused applications:

"We can start using NFTs as like you know one cent things that aren't worth very much but people attach meeting to simply because they follow a creator and was airdropped it from them."

The Rise of Hybrid NFT Standards

An interesting development in the NFT space is the emergence of hybrid standards that combine aspects of traditional NFTs with more liquid, token-like properties. Tcheyan expresses optimism about this trend:

"I hope so. I'm not going to lie I haven't looked into it enough to speak about it. Speak about it educate or smartly but I really do hope so and it's definitely something I've been wanting to look in because I think it's a I think it's an interesting new stamp."

These hybrid models could potentially address some of the liquidity issues associated with traditional NFTs while maintaining the unique properties that make NFTs attractive to collectors and creators. By combining the emotional attachment people have to unique digital assets with the ease of trading associated with fungible tokens, these new standards could drive a new wave of innovation in the NFT space.

The Future of Crypto-AI Integration

As the conversation concludes, Tcheyan emphasizes the early stage of crypto-AI integration and the potential for significant developments in the coming years:

"I think crypto and AI is very early and I think we're just beginning to see the base primitives come out and so even if you're skeptical of it right now which you should be I would continue to watch it because I think in a year or two we'll start seeing some pretty interesting products emerge."

This perspective underscores the importance of continued research, development, and experimentation in the space. As the foundational technologies mature and new use cases emerge, the intersection of cryptocurrency and artificial intelligence could lead to transformative innovations that reshape how we interact with digital systems and manage value in the digital age.

In conclusion, the conversation with Lucas Tcheyan provides a comprehensive overview of the current state and future potential of crypto-AI integration. From decentralized hardware solutions to autonomous AI agents, the possibilities are vast and largely unexplored. As projects continue to build and iterate on these foundational primitives, we may well be on the cusp of a new era in technology – one where decentralized, AI-powered systems play a central role in our digital lives. For those interested in the cutting edge of technology, the intersection of crypto and AI remains an exciting space to watch, full of challenges, opportunities, and the potential for groundbreaking innovation.

Facts + Figures

• Lucas Tcheyan is a researcher at Galaxy Digital, focusing on Solana and the intersection of crypto and AI.
• Tcheyan identifies three layers in the crypto-AI integration: hardware layer, connective tissue layer, and AI agents layer.
• The hardware layer includes projects like Akash Network, Render Network, and Golem Network, which provide decentralized GPU resources.
• Zero-Knowledge Machine Learning (ZKML) is a key development in the connective tissue layer, allowing for verifiable AI computations on-chain.
• Morpheus is mentioned as a notable project in the AI agents space, supported by industry veteran Erik Voorhees.
• BitTensor is described as using a "proof of model" approach instead of proof of work, with 32 subnets for different AI tasks.
• The Graph protocol is working on indexing blockchain data in a way that's easily readable by machine learning programs.
• Solana is transitioning from a single-client to a multi-client architecture, with the rollout of Firedancer as a significant development.
• NFT trading volume on Solana has declined, with more focus shifting to liquid token trading.
• Magic Eden topped all marketplaces in volume, with 80% from Bitcoin Ordinals and 20% from Solana.
• Compressed NFTs are mentioned as an exciting development on Solana, opening doors for new use cases.
• The emergence of hybrid NFT standards combining aspects of traditional NFTs with more liquid properties is noted as a potential trend.
• Tcheyan suggests that the true potential of crypto-AI integration may be realized in the next year or two as foundational primitives mature.

Questions Answered

What are the three layers of crypto-AI integration identified by Lucas Tcheyan?

Lucas Tcheyan identifies three main layers in the integration of cryptocurrency and artificial intelligence: the hardware layer, the connective tissue layer, and the AI agents layer. The hardware layer focuses on providing decentralized computational resources, particularly GPUs, for AI processes. The connective tissue layer bridges the gap between blockchain limitations and AI requirements, often using zero-knowledge proofs. The AI agents layer involves autonomous entities that can interact with blockchain systems to perform complex tasks on behalf of users.

How does the hardware layer in crypto-AI integration work?

The hardware layer in crypto-AI integration primarily involves creating decentralized marketplaces for GPU resources. These platforms allow individuals to rent out their unused GPU power to those who need it for AI computations. Projects like Akash Network, Render Network, and Golem Network are examples of this layer. The goal is to provide a more distributed and accessible source of computing power for AI processes, addressing issues of centralization and resource scarcity in the current AI landscape.

What is Zero-Knowledge Machine Learning (ZKML) and why is it important?

Zero-Knowledge Machine Learning (ZKML) is a technology that allows for the verification of AI computations on a blockchain without revealing sensitive data. It's an important development in the connective tissue layer of crypto-AI integration. ZKML enables the running of complex AI processes off-chain while still providing on-chain proofs of the computations' validity. This technology is crucial for maintaining privacy and security while leveraging the transparency and immutability of blockchain networks for AI applications.

How are AI agents being integrated into cryptocurrency systems?

AI agents in cryptocurrency systems are autonomous entities trained on specific datasets to perform complex tasks. These agents can have their own crypto wallets, allowing them to execute transactions, interact with DeFi protocols, or make governance decisions in DAOs. Projects like Morpheus are exploring the potential of AI agents in crypto. While still in early stages, the vision is for these agents to automate complex crypto operations and strategies based on user intents and market conditions.

What challenges do crypto-AI projects face in competing with centralized AI companies?

Crypto-AI projects face significant challenges in competing with centralized AI giants like OpenAI and Microsoft. These challenges include access to vast amounts of data, substantial financial resources, and established user bases. Additionally, the complexity of blockchain technology and the early stage of many crypto-AI primitives make it difficult to create user-friendly products that can compete with centralized alternatives. However, Tcheyan suggests that as the ecosystem matures and projects build on each other's innovations, unique and compelling use cases may emerge that centralized systems cannot easily replicate.

How is Solana positioning itself in the crypto-AI ecosystem?

Solana is positioning itself in the crypto-AI ecosystem through several key developments. It's transitioning from a single-client to a multi-client architecture, which is seen as a step towards greater decentralization and robustness. The blockchain is also focusing on technical innovations such as asynchronous execution, concurrent leaders, and economic back pressure. These advancements aim to make Solana a high-performance platform capable of supporting complex AI applications and services. Additionally, Solana's strong developer community and focus on scalability make it an attractive option for building AI-powered decentralized applications.

What is the current state of NFTs on Solana, and how is it evolving?

The NFT ecosystem on Solana has evolved from its initial focus on speculative PFP (profile picture) collections. While there has been a shift towards more liquid token trading, NFTs are not considered "dead" on Solana. Instead, there's a trend towards reimagining NFT use cases. Innovations like compressed NFTs are opening up new possibilities. There's also interest in hybrid NFT standards that combine aspects of traditional NFTs with more liquid, token-like properties. Projects like Tensor continue to innovate in the NFT space on Solana, focusing on improving liquidity and creating new utilities for NFTs beyond simple collectibles.

How might user interfaces evolve in crypto-AI applications?

User interfaces in crypto-AI applications are likely to evolve towards more intuitive and context-aware designs. While chat-based interfaces using natural language processing are being explored, the future may involve a combination of AI-powered intuitive design and context-aware systems. These interfaces could automatically adapt based on user behavior and preferences, simplifying complex crypto operations without requiring users to type out lengthy instructions. The goal is to make crypto and AI technologies more accessible to a broader audience by reducing the learning curve and improving user experience.

What is the potential impact of decentralized data marketplaces on AI development in crypto?

Decentralized data marketplaces have the potential to significantly impact AI development in the crypto space. These platforms aim to incentivize individuals and organizations to share their data, creating more diverse and comprehensive datasets for training AI models. By leveraging blockchain technology and token incentives, these marketplaces could democratize access to high-quality data, potentially leveling the playing field between decentralized projects and centralized AI giants. This could lead to the development of more robust, diverse, and innovative AI models within the crypto ecosystem, addressing one of the key advantages currently held by centralized tech companies.