Ethereum Co-founder Vitalik Buterin Urges Caution in Integrating AI with Blockchain Technology
In a blog post on January 30, Ethereum co-founder Vitalik Buterin advised developers to exercise caution when combining artificial intelligence (AI) with blockchain technology, citing potential risks and vulnerabilities.
Buterin outlined his primary concerns with AI deployment in blockchain applications, emphasizing the need for careful consideration, especially in high-value and high-risk contexts. He warned developers about the potential consequences of vulnerabilities, highlighting scenarios where a significant amount of money could be at risk, such as in prediction markets or stablecoins utilizing AI oracles.
The Ethereum co-founder acknowledged the potential viability of AI in specific use cases, such as allowing AI participation in micro-scale mechanisms like prediction markets. He also suggested the potential utility of AI in crypto wallet interfaces, aiding users in understanding transactions, detecting scams, and enhancing overall safety.
While Buterin expressed reservations about fully autonomous AI interfaces due to increased risk, he considered the integration of AI as a complement to conventional interfaces as increasingly viable.
However, Buterin deemed harnessing AI for enforcing rules or governance in crypto systems as the riskiest endeavor. He pointed out the vulnerability of open-source AI models to adversarial attacks, contrasting them with closed-source AI models that provide security through obscurity but lack transparency and assurances of unbiased operation.
Drawing attention to closed-source AI examples like the Worldcoin crypto startup, Buterin highlighted the importance of not letting anyone directly access the AI model and utilizing trusted hardware for security.
Addressing the potential challenges, Buterin identified the creation of a decentralized AI using crypto and blockchain as a significant hurdle. He discussed applications attempting to establish a decentralized trusted AI as a “singleton” for specific purposes, acknowledging their potential but warning of underlying assumptions that could fail.