Crypto and AI: Just Scratching the Surface, Says Microsoft Exec

Microsoft views artificial intelligence (AI) as “the defining technology of our time” and has been leading the way in both AI research and investment. This doesn’t mean the tech giant isn’t also tracking developments in the world of cryptocurrencies, exploring how blockchain technology and AI could one day support each other. At the recent Cornell Blockchain Conference in New York, Yorke Rhodes, Microsoft’s director for digital transformation, blockchain, and cloud supply chain, addressed this potential intersection. He mentioned that as AI and blockchain technologies advance, there might be avenues for creating systems that harness the strengths of both, but we’re only beginning to scratch the surface.

During a panel titled “Crypto x AI,” Alex Lin, co-founder and general partner at Reforge, questioned Rhodes on whether Microsoft would ever develop its own blockchain. Rhodes pointed out that the current crypto landscape is filled with innovative projects, especially within the open-source community. He suggested that Microsoft is more inclined to optimize existing technologies, such as layer-2 blockchain rollups, rather than creating a new blockchain from scratch.

The discussion featured other prominent figures too, including Neil DeSilva, CFO at PayPal Digital Currencies; Matt Stephenson, head of research at Pantera; and Jasper Zhang, CEO and co-founder at Hyperbolic Labs. Stephenson stated that crypto may serve as the “picks and shovels” for certain types of AI, especially transformer and diffusion models. Despite the potential, AI seems to dominate the spotlight, often overshadowing other emerging technologies like crypto/blockchain and Web3.

Lin remarked on the buzz surrounding the merger of blockchain networks and AI but cautioned about separating hype from reality. He noted that while decentralized graphics processing units (GPUs) are garnering attention, issues like latency—the time taken for data to move across networks—are often overlooked. According to Lin, centralized AI systems currently offer quicker responses than their decentralized counterparts due to latency issues.

Contrary to Lin’s concerns, Zhang from Hyperbolic Labs didn’t see latency as a significant problem for decentralized networks. He explained that decentralized systems could drastically reduce communication delays by processing requests locally. For instance, a user in London could send a request to a nearby node rather than a distant data center, resulting in quicker processing times. Hyperbolic Labs successfully launched an AI inference interface that matched centralized solutions in terms of latency, proving that decentralized networks could indeed keep pace.

The trend in AI discussions often gravitates towards large language models (LLMs) that demand extensive computing power. Yet, Rhodes emphasized the rise of “edge AI,” which involves smaller language models that can run efficiently on devices like smartphones and laptops. Smaller models have become more practical because there is more computational power available on edge devices now. Microsoft has been focusing on developing these smaller language models, such as its Phi-3 family, which require less data and computational effort to function.

The regulatory landscape for AI is likely to become stringent, similar to what cryptocurrencies have experienced. Panelists discussed the challenges of regulation, particularly in the context of AI’s decision-making opacity. Lin pointed out that the U.S. Securities and Exchange Commission (SEC) plans to regulate AI even more rigorously than digital assets. DeSilva from PayPal expressed a different view, arguing that while U.S. regulations can be challenging, they are generally aimed at protecting consumers.

Despite potential regulatory hurdles in the U.S., other regions like the European Union are becoming more welcoming to innovative fintech solutions, including stablecoins. DeSilva cautioned that if the U.S. doesn’t accelerate its regulatory pace, it could lose its competitive edge. He suggested that AI’s growth could benefit from blockchain’s transparency and tracking capabilities, potentially easing regulatory concerns.

The session concluded with insights into AI’s future. Zhang predicted that AI would become powerful enough to be universally adopted, transforming every company into an AI-driven enterprise. He forecasted that AGI (artificial general intelligence) might become a reality in the next five to ten years, enabled by decentralized infrastructures that can aggregate computational power. Rhodes added that zero-knowledge proofs might be replaced by fully homomorphic encryption within three years, offering greater privacy and security, especially in sensitive fields like healthcare.

DeSilva, reflecting on decades of experience, mentioned that optimism often leads to positive outcomes. He expressed a hopeful view that AGI would eventually be realized and that it would benefit humanity, provided that everyone’s collaborative efforts are directed towards this goal.