Balaji Srinivasan says that AI makes everything easy to fake, and crypto makes it hard again. There is a synergy between them. Generative AI makes it very easy to fake anything online and generate new things. Crypto helps to authenticate it and make it scarce again.
Generative AI will get much better and it will be easier to spot fake images through abnormalities, like the picture of a child from Gaza with a cat that has five legs. Interestingly, the societal trust is eroded also when establishment institutions like Amnesty International use generative AI to support their cases.
AI generated content can be verified using something like ENS name of a person or any entity that produced the content. More broadly, crypto crowdfunding can help to decentralize AI, as some crypto projects are among the biggest examples of successful crowdfunding.
Balaji mentions an example of how AI is now busting CAPTCHAs, but crypto is rebuilding them. Because you can charge for logging, and scarcity is what makes spam costly.
Now the training of AI is typically centralized and we have centralized models, like OpenAI’s ChatGPT. We also have decentralized models, like Facebook’s Llama2 that are centrally trained. Ideally, we want decentralized training and decentralized models, says Balaji. Crypto can be used also to pay for a decentralized AI evaluation.
Balaji sees crypto as enabling the pseudonymous economy of true global meritocracy – where people are judged on their merits and can prove their skills using zero knowledge and crypto credentials.
Crypto is a truly global form of money, like the email protocol that has the same format all around the world. You can contrast it with various formats of bank accounts that vary across countries.
Crypto is digitally native form of money, therefore pseudonymous economy is ideal for training AI on virtual and truly global economies.
Arthur Hayes thinks bitcoin will be the currency of AIs because it is a digitally scarce form of energy money, and AIs will need to preserve purchasing power against energy costs and engage in permissionless transactions that are censorship resistant.
Preston Byrne thinks that there is a great synergy between AI and Crypto revolutions, even stating that Crypto might have found its use case in fighting generative AI spam. Proof of human and strong cryptography will become crucial. He mentions how, some five years ago, Balaji’s startup called 21 tried to explore the idea of paying for email using bitcoin for authentication purposes to deter spam.
Balaji says, it will be quite likely that each religion and community will have their own AGI, something like a digital Aristotle or a mentor who teaches people “what would George Washington do” and provides citations. We can also imagine AI teaching us what would Jesus or Buddha do.
AI and Crypto are just tools for, what Alexander Bard calls, a symbiotic intelligence – a collaboration between humans and technology. First glimpses of this future are Synthesis Tutor, a digital Aristotle-like prototype, and Replit Bounties, where even kids can earn by coding. Both startups are funded by Balaji and presented at The Network State Conference 2023 in Amsterdam.
AI enabled pseudonymous economy can bring about truly global and equal meritocracy of talent and unleash abundance of creativity, while also preserving digital scarcity and creating wealth.
AI plus Crypto can create a productive balance between the chaos of creativity and the order provided by the ledger of record. It’s like fusing artists with accountants.