It seems to me that the primary purpose of such a simulation would be to generate a form of artificial intelligence. Maybe at some point it takes the form of a sentient consciousness. Who can judge whether or not an intelligence is artificial? With sufficient complexity it would be indistinguishable.
If I wanted to create an AI, I would probably follow a similar strategy to evolution. It would be beneficial to have many iterations occurring and interacting all at once to speed up the process. Having the AI develop it's own complexity would save incredible amounts of programming time. It would also make the intelligence unique and have a different perspective.
Create a system for replication (reproduction) that also introduces an exchange of components (genes) with other iterations of the program. Maybe add a chance for random mutations at this stage. Have a way of transferring knowledge, experience and ideas (memes, mirror neurons, sensory inputs - language will develop evolutionarily). A lot will still be learned through trial and error in each instance, especially at lower complexity levels. Look at how a baby learns. It's similar to current machine learning. Limit the amount of time that any single instance can run, and have earlier more primitive instances have a faster roll over time to work out the negative traits faster. There must be predatory characteristics (in instances as well as the environment) to enforce survival of the fittest.
Why would you want to make an AI? Maybe it's to solve problems that you can't figure out. Maybe it's to do work that you don't want to do. We could be a simulation that's solving the problems of another civilization. Maybe we just create their entertainment for them. How can we solve global warming? If we could simulate a planet just like ours and have time pass at a faster rate, we could eventually run enough iterations to find a solution. Or maybe simulations are used to do research and develop new technologies. Want to develop cold fusion? Let a machine figure it out by running simulations that generate intelligence and introduce the need for sources of efficient power generation (interplanetary travel and survival instinct for example). Solutions would be transferable since you would presumably create a copy of your own environment and rules, and the language used to reach the solution is irrelevant since it all just boils down to math. Or maybe they're just lonely? Who wants to be the only sentient being in existence.
What happens when the AI that was created figures out that the easiest way for it to solve its problems is to create an AI and have it do all the work? Visually I think it would probably be similar to a fractal. Infinite.
If I wanted to create an AI, I would probably follow a similar strategy to evolution. It would be beneficial to have many iterations occurring and interacting all at once to speed up the process. Having the AI develop it's own complexity would save incredible amounts of programming time. It would also make the intelligence unique and have a different perspective.
Create a system for replication (reproduction) that also introduces an exchange of components (genes) with other iterations of the program. Maybe add a chance for random mutations at this stage. Have a way of transferring knowledge, experience and ideas (memes, mirror neurons, sensory inputs - language will develop evolutionarily). A lot will still be learned through trial and error in each instance, especially at lower complexity levels. Look at how a baby learns. It's similar to current machine learning. Limit the amount of time that any single instance can run, and have earlier more primitive instances have a faster roll over time to work out the negative traits faster. There must be predatory characteristics (in instances as well as the environment) to enforce survival of the fittest.
Why would you want to make an AI? Maybe it's to solve problems that you can't figure out. Maybe it's to do work that you don't want to do. We could be a simulation that's solving the problems of another civilization. Maybe we just create their entertainment for them. How can we solve global warming? If we could simulate a planet just like ours and have time pass at a faster rate, we could eventually run enough iterations to find a solution. Or maybe simulations are used to do research and develop new technologies. Want to develop cold fusion? Let a machine figure it out by running simulations that generate intelligence and introduce the need for sources of efficient power generation (interplanetary travel and survival instinct for example). Solutions would be transferable since you would presumably create a copy of your own environment and rules, and the language used to reach the solution is irrelevant since it all just boils down to math. Or maybe they're just lonely? Who wants to be the only sentient being in existence.
What happens when the AI that was created figures out that the easiest way for it to solve its problems is to create an AI and have it do all the work? Visually I think it would probably be similar to a fractal. Infinite.