Naval Ravikant on Naval
Live in the Future
Nearly all of this material is duplicated in other supplied versions of “Live in the Future,” with additional overlap with “The AI Industrial Revolution.” Skip unless you specifically want the clearest discussion of machine-to-machine language and AI legibility.
not legible by the way. It's an anachronism. Our training data is in natural language and in computer languages that humans can read and write. And so the AIs are trained on that. But in the future the AIs will develop their own languages. They'll communicate with each other in some kind of a binary high speed protocol. They'll almost be talking compiled code or maybe even something completely different that we can't even imagine which will be more efficient for communication. It's only humans who demand the legibility, but the AIs don't really have it. And part of what makes AI systems so much better than human coded expert systems is that they don't have to be legible. So they can find correlations and patterns that we cannot articulate. In fact, even most of our knowledge is inarticulatable. Most of your feelings are inarticulatable. You don't even know half the time what you're feeling or thinking.
Naval discusses the speed and societal consequences of AI progress, including questions about creativity, governance, open source, and the limits of human oversight. He also offers a more optimistic view in which people use AI and robots as leverage, remaining valuable through judgment, guidance, and desire.