04-06-2017, 09:58 PM
On a long enough timeline, AI will likely be able to do ANYTHING humans can do. As well, or better. An entity that can learn and develop at an exponential rate will only be limited by input, i.e. how well can it understand humans? And at what resolution? Will it only have access to, say, wikipedia type stuff or ALL of the internet and everyone posting there?
With a high enough input resolution [understanding of human behavior] I think AI will definitely be able to break the rules in exactly the same ways that humans can. Because wanting to break the rules to create something different is a clearly discernible pattern in human behavior. And I would think that an AI eats discernible patterns for breakfast. Coming up with logic that injects semi-random "breaking the rules like humans do" things into an otherwise algorithmic music creation software isn't going to be that big of a deal for a system that advanced. And we wouldn't hear any difference.
With a high enough input resolution [understanding of human behavior] I think AI will definitely be able to break the rules in exactly the same ways that humans can. Because wanting to break the rules to create something different is a clearly discernible pattern in human behavior. And I would think that an AI eats discernible patterns for breakfast. Coming up with logic that injects semi-random "breaking the rules like humans do" things into an otherwise algorithmic music creation software isn't going to be that big of a deal for a system that advanced. And we wouldn't hear any difference.