Restoring Authenticity in the Digital World

One intriguing concept is the fusion of AI and cryptography. We often hear these two terms paired together, but I believe there’s a specific overlap in their technologies. The key here is that while AI, particularly generative AI, makes it easy to fake anything online, cryptography verifies and authenticates the data, restoring its scarcity.

For instance, take an AI-generated arrest photo of Donald Trump. The current advice to verify its authenticity is to scrutinize the fingers, as generative AI hasn’t perfected its rendering. However, as technology progresses, this method will become obsolete. The solution lies in cryptographic verification, such as digital signatures tied to an Ethereum Name Service (ENS) address.

Much of this infrastructure is already in place, using ENS and IPFS. However, we need clear examples of content verified in this way. This is crucial because once we have a reliable origin or citation, we can determine whether the content was human-generated or AI-generated.

Now, you might argue that only a few things can be proven on-chain, such as financial records within the crypto ecosystem. This is akin to the early days of the internet, where content was limited. But as Web 3.0 advances, we’re seeing more data being put on-chain, not just financial transactions but also NFTs and social interactions.

The aim here is to build a decentralized web three of trust, a network where we can verify the actions of entities through their public keys and interactions. This is a departure from the current internet model, where content is ranked by Google’s page rank, which increasingly ranks fake content.

Crypto also can potentially rebuild Captcha systems, which AI is currently cracking. By requiring small payments or a certain staked amount of money, we can make spamming costly.

Another significant aspect of crypto is the decentralization of AI training. Currently, AI training is mostly centralized, with companies like OpenAI leading the way. However, with the advent of crowdfunding via crypto, we can potentially decentralize the process.

In addition to decentralized training, there’s also the potential for decentralized evaluation. Just as there are new releases of Ethereum or Solana that people update on their nodes, we could see updated versions of models that people host.

Finally, there’s the concept of a polytheistic AGI (Artificial General Intelligence), where every community has its own Oracle, or AI model, that they consult. This model could give on-chain citations, pulling together many concepts discussed.

In conclusion, while AI has made it easy to fake online content, crypto can potentially restore authenticity.