Sam talks to Teng Yan: Top ten insights on Crypto x AI

This episode of the "Hack and Tell" podcast is hosted by Sam Williams, featuring Teng Yan as a guest. They delve into the future of Crypto, AI, and Decentralized Computing (AO).

Background Information

• Sam Williams (@samecwilliams): Founder and inventor of Arweave and AO, also serves as the CEO of Forward Research.
• Teng Yan (@0xPrismatic): Former researcher at Delphi Digital with over 10 years of experience in the tech industry. Currently heads the independent research company Chain of Thought, focusing on the intersection of crypto and artificial intelligence.

Key Insights

1. Synergy Between Crypto and AI

Both Teng Yan and Sam believe that there is a lot of hype in the field of combining Crypto and AI. However, this integration brings real disruptive potential, especially in building trustless autonomous systems and intelligent financial agents.

Teng Yan points out that the current hype and experimentation indicate that the industry is rapidly evolving, while Sam emphasizes that intelligent financial agents, as a trusted protocol, can operate without the need for trusted intermediaries, which is crucial.

2. Origins of AO

Sam mentions that AO was initially designed to address the content trading needs of decentralized social media networks (Odysee, which has around 7-8 million monthly active users and has now been acquired by Forward Research). The team ultimately developed a new blockchain supporting scalable smart contracts based on this need to create a system that supports content ownership trading and efficient markets, which was an innovative outcome triggered by solving their own problems.

3. Transformation of Agent Finance

Sam describes agent finance as the most intuitive combination of decentralized AI and crypto technology. He notes that decentralized financial agents can eliminate reliance on traditional intermediaries (such as funds and banks) and allow ordinary developers to participate in creating financial strategies. This reduction in trust and lowering of entry barriers will unleash a significant amount of untapped innovative potential.

Teng Yan believes that most transactions on the blockchain are currently finance-related, making agent finance the best entry point for AI agents into the blockchain.

4. AI Technological Progress on AO

On a technical level, Sam shares how AO supports LLMs (like Llama.cpp) running in smart contracts, emphasizing that this is an important milestone for bringing AI into the blockchain. He acknowledges that performance is still lacking (processing 0.7 tokens per second), but this capability provides a solid foundation for future development of complex AI intelligence.

Teng Yan states that although the technology is not yet mature, these experiments have paved the way for developing AI-driven practical applications.

5. Practical Applications of AI on AO

Sam details the case of Llama Land within the AO ecosystem, which is an experimental platform allowing users to experience decentralized governance and monetary policy through an AI-managed autonomous system. The core of Llama Land is an AI (Llama 3) capable of autonomously deciding on monetary issuance, where users can submit requests, and the AI automatically decides whether to grant rewards based on those requests. This is not only an interesting gameplay but also showcases the potential of autonomous AI in on-chain economies.

6. Philosophical Discussion on AI

Both express similar sentiments—the rise of decentralized AI is not just a technological innovation but also a philosophical debate between open-source and centralized models.

Teng Yan believes that the open system of decentralized AI allows global developers and users to participate, breaking the monopoly of large tech companies and making AI development more democratic.

Sam emphasizes that the neutrality and trustworthiness of execution in decentralized AI are fundamentally different from traditional centralized AI. This open system acts like a protocol that can execute in a neutral manner, free from control or manipulation by any single entity.

7. Building AI Infrastructure

Sam points out that the current focus is on establishing infrastructure that can attract developers, including tools and frameworks, to quickly build and deploy AI applications. By lowering the barriers to innovation, developers can more easily realize the cycle from learning to development to production.

Teng Yan adds that AI agents on the blockchain represent an ecosystem that develops gradually from underlying infrastructure to applications, requiring good development tools and foundational platforms to support it.

8. Market Behavior and Growth

Teng Yan mentions that the emotional fluctuations in the crypto market are often unrelated to technological progress, yet many people view price as an indicator of technological maturity, which is a misunderstanding.

Sam adds that true technological development is gradual, and it is common for the market to overestimate short-term potential while underestimating long-term impacts. He also observes that despite fluctuating market sentiments, technological capabilities are steadily improving.

9. Vision for the Future of AI

Teng Yan and Sam agree that AI-driven blockchain applications will lead to exponential growth.

Teng Yan is more optimistic, believing that this growth will rapidly manifest in the coming months.

Sam believes that we will continue to feel the profound impacts of this change over the next 10 to 20 years. From financial agents to AI autonomous governance, these technologies will gradually permeate more industries and applications.

10. Looking Ahead to 2025 and Beyond

Teng Yan predicts that 2025 could be a key milestone in the AI field, as technological infrastructure will gradually improve in the coming months, and large-scale developer participation will trigger rapid application expansion.

Sam points out from a long-term perspective that the combination of AI and blockchain is a lengthy and profound process, with impacts that may last for decades, but the emergence of early users and use cases will significantly accelerate this process.