What Vitalik Expects from AI + Crypto

AI + Crypto is a topic we’ve spent a good amount of time with here at Bankless. It seems to be what Vitalik has been thinking about lately, too.

His latest article discusses the potential intersections between these two technologies. Today, we’re giving you the tl;dr, but as always, the full thing is worth a read! In his piece, Vitalik spends most of his time categorizing where AI could find utility inside crypto (.i.e., the Game), focusing on four broad areas to discuss their promises and challenges:

1. AI as a Player in a Game:

Vitalik sees this as highly viable — AI agents engaging in blockchain transactions, such as DEX swapping or arbitrage trading. A prime example he cites is AIOmen, an AI-driven prediction market. In these markets, AIs provide a powerful tool that makes lots of small, smart guesses, improving how well these markets work and the quality of information we get from them. For example, even if a question in the market isn’t interesting to most people, AIs will still work on it since they aim for overall accuracy in their predictions, not single big wins — improving the overall quality of the market.

2. AI as an Interface to the Game:

High potential, yet risky, says Vitalik. AI could guide users through crypto, offering features like Metamask’s scam detection. AIs could explain the inner workings of complex transactions, enhancing user understanding and security. However, significant risk arises if people can manipulate AI defenses, using them to exploit instead of explain.

3. AI as the Rules of the Game:

The riskiest category. It involves AIs enforcing rules of smart contracts or DAOs — akin to digital judges. Again, the risk here lies in whether attackers can exploit AI vulnerabilities, especially if the model or its training data is transparent.

4. AI as the Objective of the Game:

More long-term, this category involves using blockchains for developing and maintaining AIs to do things like training and data protection. Vitalik asserts decentralizing AI this way could contribute to AI safety by creating natural kill switches for agents.

The past year of generative AI progress has pushed investors to rethink how the field could remake the tech economy, and it’s led plenty of investors to drop money into fledgling AI coins. The reality is that most early efforts today to integrate AI into crypto are barely scratching the surface.

Overall, Vitalik is clearly excited by these possibilities for AI + crypto, though he definitely seems to believe these innovations must be approached cautiously – something he talked about in depth a few months ago in another blog post.