AI agents handling money need blockchain's security, not direct bank access. The speaker explains how proof-of-work and distributed networks make it nearly impossible for an AI to steal or go rogue, since compromising Bitcoin or Ethereum would require controlling 51% of the network's computing power.
You don't want to attach an AI agent to your bank account. It could drain your bank account and it could destroy the economy, bro. They need to be contained in an unhackable environment. Why? Because there's a massive problem. So if you tell, for example, the AI to be like, hey, go buy these pair of shoes for $100 and only spend $100. Before blockchain, it goes rogue. With blockchain, because of cryptographic technologies, essentially you cannot hack a blockchain. There's a concept called antifragility. So as the network gets bigger, it gets harder to hack. In Bitcoin, proof of work is this: a whole bunch of computers pointed at solving the block. So there's a block of transactions. I want you to imagine a ledger. A ledger is basically like you go to a gym and you write your name and there's a list of names. That's a ledger. Okay? A ledger that has transactions on it—that's the blockchain ledger. So every ten to fifteen minutes, that list gets updated. And imagine billions and billions of dollars of computational power all pointed at this one problem, which is we need to make sure. Those transactions are legit. Technically, the AI could hack it, right? AI is super smart. Well, I'm telling you it can't, because the only way it could hack it is if it somehow generates 51% of all of the computational power. That's the innovation of cryptographic technologies. So as you can imagine, as it expands, the chances of Ethereum and Bitcoin getting hacked, they lessen.
