Vitalik Buterin cautions developers to exercise caution when combining cryptocurrency and artificial intelligence.

Ethereum co-founder Vitalik Buterin has shown increased interest in the potential convergence of artificial intelligence (AI) and cryptocurrency. However, he advises developers to approach this intersection cautiously.

In the past year, AI has garnered attention in Silicon Valley, diverting focus from cryptocurrencies to large language models and platforms like ChatGPT within the realm of Web3 and the Metaverse. Although some blockchain projects have sought to capitalize on the rising AI trend, a significant number of AI-infused crypto initiatives appear to be more hype than substance.

Despite potential challenges, Buterin believes in the promise of the synergy between crypto and AI. In a blog post released on Tuesday, he shared his insights on where these two technologies might intersect in the coming years, emphasizing the need for careful consideration.

Buterin employs the analogy of a game to categorize potential overlaps between AI and blockchain into four distinct categories. The most viable category involves applications where AI serves as a player in the game. In this scenario, AI predicts outcomes in events like prediction markets, with blockchain enforcing rewards or penalties based on the accuracy of the predictions.

The second category, deemed high potential but with high risks, features applications where AI acts as an interface to the game. Examples include scam-detection features in crypto wallets, enhanced by AI’s detection and explanatory capabilities.

The third category introduces apps where AI dictates the rules of the game, essentially serving as ‘AI judges.’ Buterin warns of the need for caution when exploring this area, suggesting potential use cases in helping decentralized autonomous organizations (DAOs) make subjective decisions using AI.

The fourth category explores use cases where AI becomes the objective of the game. This involves leveraging blockchains as infrastructure to enhance the development of AI models.

While expressing increased optimism about the intersections between AI and crypto, Buterin acknowledges potential challenges in balancing the transparency inherent in crypto with the customary opacity of black box AI systems. He notes the inherent tension between the cryptographic principle of open source for security and the vulnerability to adversarial machine learning attacks when AI models or their training data are openly accessible.