2

Having read many articles and books dealing with a possible technological singularity, and also other articles and books about the possibility that we are living inside a simulation, I have been pondering about the possible consequences of both theories being correct.

The question that is now keeping me occupied is:

"When a technological singularity occurs within a simulated (closed) environment, could the technology arising from that point on ever directly manipulate the environment outside of its own simulation?"

My reasoning:

We humans, as intelligent beings, invent technology that we simulate by outfitting it with Artificial Intelligence. Once we reach the point that the AI becomes self-aware and more intelligent than us, it will start inventing technology and AI that are quickly beyond our comprehension.

The simulation that we started has at this point become superior to us and it can be argued that it will effect us directly. If not psychologically, then biologically by means of enhancements to our DNA, etc. Therefore we can say that this simulation directly effects the simulator and its environment outside of its simulation.

I therefore reason that this will also occur in the simulation of our universe, assuming it is a simulated environment. Our post-singularity technology, including post-humans, will be able to evolve beyond the capacity of our simulator and directly effect it.

John Willemse
  • 121
  • 3
  • This seems incredibly speculative (asking for a mostly-opinion response) -- any chance you might be able to specify the concern a bit more concretely? What might have made this an interesting or important concern to you? What have you found out so far? What might you be expecting in an answer? – Joseph Weissman Feb 15 '13 at 15:57
  • Simulating all that is, requires more than all that is. –  Feb 15 '13 at 16:29
  • (Just in passing, Greg Egan's Crystal Nights might be an interesting resource here.) – Joseph Weissman Feb 15 '13 at 18:17
  • @JosephWeissman I'm just a layman-philosopher, and I like to think about these sort of things. Thanks for the book reference, I will definitely check it out! – John Willemse Feb 15 '13 at 22:10

1 Answers1

2

Whether it can or not depends entirely on the physics of the world in which the simulation is being run, and the physics used to create the simulator. What other answer could there possibly be?

In particular, we know that the answer can be no it cannot because we can't distinguish our reality from something Turing-computable, and a Turing machine as described needn't affect anything outside that machine.

And we know that the answer can be yes it can because of course you can have arbitrarily much coupling between the contents of your computation and the rest of reality. (We mostly only couple heat production because of the architecture of the computers we use.)

Now, if you mean "can we use results from a simulator to impact the real world", then the answer is yes of course, that's why we build them! For example, almost all aspects of aircraft design are done in simulators before any real component is built. If you mean "should we just simulate reality and then take the answers that simulated humans (or whatever) develop and use them", then the answer is no of course not, how wasteful and error-prone because you could put your computational resources towards actually generating the answers you want instead of simulating a bunch of people playing Angry Birds, and anyway, you've probably got your simulated physics wrong and so the answers there won't translate.

(This is not really a philosophy question; it's a speculative physics question.)

Rex Kerr
  • 15,970
  • 1
  • 23
  • 46
  • Just out of curiousity -- considering that finitistic approximations that are often required for computability, is there not (perhaps in principle) the possibility we could distinguish our reality from something built entirely from Turing-computable functions? – Joseph Weissman Feb 15 '13 at 20:33
  • @JosephWeissman - No, because you can form arbitrarily precise approximations with a Turing-computable system. If we had a way to make infinitely accurate measurements we would be able to distinguish our reality, but there does not seem to be any way to accomplish this (any fanciful ideas we might have had pre-QM have been thoroughly squashed by the uncertainty principle). – Rex Kerr Feb 15 '13 at 21:52
  • One thought behind the question was that if, and only if, we live in a simulated environment, then we may be able to prove that when we hit a certain upper limit in computing power or intelligence. The reason being that inside our simulator we can never exceed the computational power of the simulator itself. Once we hit a limit and cannot explain why technology, that should theoretically cross that limit, simply does not. – John Willemse Feb 15 '13 at 22:14
  • if we find ourselves unable to cross a limit, then either (a) the theory that fails to predict it is incorrect or (b) we aren't trying hard enough. – artm Feb 15 '13 at 22:30
  • 2
    @JohnWillemse - You seem to have a very dim view of the capability of the entities creating this amazing simulation environment. Current computers when overtaxed just slow everything down--the simulation runs slower but from the inside, nothing can tell the difference. There is no reason to expect anything different. Granted, it could be implemented like some multiplayer games where more shortcuts are taken when the load is heavy, and if we find anything like that (physics gets weird when we try to compute too much), it could be suggestive. But there's no reason to expect it. – Rex Kerr Feb 15 '13 at 23:01