Was having a think about what it would look like if this world / universe were designed by an AI. It's highly likely that we will be superseded soon by our own creations ergo seems sensible that if any Godlike being arises from biological processes it will be constructed rather than born. So what evidence would such an AI leave?
Well here goes, caveat being that we are at the root of our AI so it reflects the limits of or thinking - in particular Western thinking.
I like playing PC games which is often at the forefront of AI thinking. It struck me that the strength of AI is its overall ability to accurately gather information , weigh alternatives and consider solutions within the confines of its programming. Any game you play now has the rules 'cooked' such that the AI always produces perfect solutions - it doesn't experience an emotional response. An AI world (I'm talking about the environments it creates not its ability to learn) repetitively generates itself but without feedback variation (i.e. its not evolutionary, its repetition according to parameters) so when you see a tree it is Platonic, it is a modified representation of an ideal tree (the one in memory it is cloning) and so on. AI learning is not subject to forgetfulness though it is limited by storage and retrieval considerations. AI always has an aim, so a learning system will always be trying to learn for a specific reason (just like the current Go contest- the AI is learning to win that game not learning to understand love or some abstract concept.)
Ok so assuming such an AI was able to generate a universe using the application of some physics as yet undiscovered or conversely run a simulation that exhibited physics like rules. Would such an AI be evident and what sort of evidences might it leave? Why would an AI generate a universe filled with biological systems rather than one filled with AI robots? Why would a biological system/simulation, given enough time, end up regenerating an AI?
- I think an AI would produce perfect scenarios to reach its goal (the AlphaGo AI doesn't bother generating a play or musing on the taste of coffee to better itself at Go, an AI with a purpose would generate a world closer to scripture i.e. the Garden of Eden story sounds exactly like the sort of world an AI might produce, one where a set of conditions is suggested, all other variables are held invariant and the simulation is run with limited parameters leading to a conclusion.) The messiness and pointlessness of actual reality argues against any kind of scenario / test whereas the artificiality of scriptural stories would support a perfect scenario approach.
- An AI could run multiple simulations at the same time or multiple physical universes side by side. Science has branches of maths that would support this.
- An AI would be likely to be based from mathematical principles. Some parts of the universe support this (that maths itself can approximate the universe suggests that there are underlying principles that govern the universe that can be abstracted to mathematics) but then some parts of the universe don't map well with maths - maths can describe impossible scenarios such as infinity.