Vitalik writes on the promise and challenges of crypto + AI

True to Vitalik’s style, it is a long, technical, and dense read – but deeply insightful. 

So, we’ve read through it, so you don’t have to.

Here is the lowdown on how Vitalik sees AI and crypto intersecting. 

Vitalik categorises crypto+AI applications into four major types:

AI as a player in a game

The most viable option. AI bots can participate in blockchain-powered games (not necessarily gaming, more like game theory interactions), markets, and mechanisms as automated players. We've already seen how arbitrage bots on DEXs and prediction markets populated by AIs have been revolutionary. Autonomous agents are the viable next step, and it would be interesting to see how things shape out when they transact and credibly commit to agreements. 

Interestingly, we are bullish on autonomous agents and have published a report on them – make sure to check out our report on Crypto x AI.

AI as an interface to a game 

According to Vitalik, this use case holds potential, but there are significant risks around adversarial attacks. AI could help users understand activities on-chain, explain the consequences of transactions in plain language, and detect scams. However, publicly exposing AI assistants also exposes vulnerabilities. A balanced approach is needed.

AI as the rules of the game

This is where we need to tread very carefully. While this use case is the most exciting, Vitalik argues it is also the most risky.

Using AI to make judgments or decisions directly introduces efficiency and security issues. Adversarial attacks on public models and high computational overhead for encrypted models pose barriers. 

Approaches like delayed publication of rankings can improve centralised AI governance. However, we are still a long way from reducing the risk for this use case to be adopted in mission-critical applications. In Vitalik’s words, “a prediction market or a stablecoin that uses an AI oracle, and it turns out that the oracle is attackable, that's a huge amount of money that could disappear in an instant.”

AI as the objective of the game

Vitalik sees a longer-term potential in this use case. Blockchains and cryptography may enable the development of unbiased, privacy-preserving models safe from misuse. This could expand the utility of AI while avoiding the centralisation risks of mainstream approaches.

Cryptonary’s take

Yes, there are technical hurdles at the intersection of crypto and AI. Adversarial attacks remain a threat; closed models risk centralisation, while open models allow exploitation. In addition to the technical hurdles, there’s also the social dynamic – and, of course, regulatory challenges. 

But beyond these hurdles, there’s no denying that AI could simplify and protect users’ Web3 experiences. 

If cryptographic techniques can scale, they could produce AI systems whose objectives and decisions earn public trust. As Vitalik rightly points out, the path forward isn't straightforward, but the possibilities make it well worth exploring. 

Here at Cryptonary, we see promising potential in applications at the intersection of crypto and AI. Our 2024 predictions affirmed that AI and crypto are complementary technologies that will positively reinforce each other. 

Our latest research into the world of autonomous agents makes a case for betting on an industry still in its infancy. 

Will you be bet on the future?