-2
$\begingroup$

Assume an average AI situated in an environment with enough computing power, like a Matrioshka brain.

The AI does not interact directly with the simulations.

There are some explanations like a curious AI or a scientific AI, but I'm curious what you could come with. Hard sci-fi would be appreciated. I'm striving for a semi fermi paradox solution.

EDIT: The universes are in name only, the AI is just providing a believable environment for its occupants. The occupants are its target, not the universe.

$\endgroup$

closed as primarily opinion-based by JBH, elemtilas, Ender Look, L.Dutch Jul 24 '18 at 4:10

Many good questions generate some degree of opinion based on expert experience, but answers to this question will tend to be almost entirely based on opinions, rather than facts, references, or specific expertise. If this question can be reworded to fit the rules in the help center, please edit the question.

  • 1
    $\begingroup$ Welcome to Worldbuilding.SE! We're glad you could join us! When you have a moment, please click here to learn more about our culture and take our tour. You have asked what we call a high concept question and these questions tend not be answered well without improvement. Please follow the links for details. $\endgroup$ – JBH Jul 24 '18 at 0:15
  • 1
    $\begingroup$ Your question needs improvement. (a) explain to the crowd what a Matrioshka Brain is (never assume we'll just go look it up). (b) What do you mean by a semi-Fermi Paradox? How does that relate to your answer? (c) What's the point of simulating multiple universes if they're only window-dressing for the species? Can the species envision/travel-to these mutliple universes? Is the universe simulation only of a star map in the sky? (d) What's the purpose of the simulation? and (e) what constitutes a good answer? How will you judge the results? $\endgroup$ – JBH Jul 24 '18 at 0:34
  • $\begingroup$ Also, your edit contradicts your question: if the AI only needs to "provide a believable environment" for the occupants, then it doesn't actually need to simulate multiple universes, which, as you say are "in name only". $\endgroup$ – elemtilas Jul 24 '18 at 0:37
  • $\begingroup$ I apologize for being erratic with the edits and the format. I did not know of the question sandbox existence. To prevent any more confusion I am marking this question as solved. Thanks for JBH's help, it's appreciated. $\endgroup$ – Fargutin Jul 24 '18 at 1:06
  • $\begingroup$ Why you wrote [solved] on your question's title?? Please delete that. $\endgroup$ – Ender Look Jul 24 '18 at 1:45
2
$\begingroup$

Essentially, the AI is looking for a particular emergent property, so it runs parallel simulations, each one with the same "species", but changing the parameters of the various simulation universes.

While interesting in of itself, the AI is likely doing this because it intends to seed the real universe with this species, in order to harvest whatever the effect of this emergent property actually is. For a truly awesome SFNal property, condsider "Childhood's End", by Arthur C Clarke. While the Overlords may have almost god-like powers, the Overmind wants the essence of Humanity, and dispatches the Overlords to shepherd us until such time as we have reached that level of consciousness.

Since there is no "Overlords" or Humanity, the AI must work out the possible parameters. Based on the information of the OP, the AI seems to have the ability to create both the species and set up the civilization, and create a "pocket universe" in order to run the experiment in realtime once it has the correct parameters established.

Alternatively, and more simply, the AI leaves the parameters of the Universe the same, and makes subtle changes to the simulated species until it gets the results it wants. As before, thousands or millions of runs in parallel are done to find the correct answer.

$\endgroup$
1
$\begingroup$
  • It doesn't need more than one civilization in order to complete its investigation.
  • It doesn't have enough computational power to simulate several civilizations so it will only simulate one at the same time.
  • It isn't sure if the simulation will work, so it will try first with only one civilization, and if the project is a success, it will process with several civilizations.
$\endgroup$
1
$\begingroup$

The way I read the question is, the AI wants to (is programmed for?) achieve some kind of result with the observed species.

But it is limited in its capability of modeling the species itself - it can't "solve" its equivalent of a psychohistoric equation set, and conclude that it needs a universe made just so. It can only simulate the species and its reactions to the environment in order to observe its emergent properties, therefore the AI proceeds by trial and error, by adjusting the "universe" parameters around the observed species. Mostly I'd expect it to manipulate the simulation of the species' home solar system - climate, duration of the year, composition, structure and so on. Let them see lots of stars, a few stars, no stars at all; things like a nearby Moon to stimulate space exploration, or periodical, dangerous meteor showers.

I seem to remember something of the kind happening with Ambrotos in Sawyer's Terminal Experiment (there, a species was designed from the ground up to achieve a specific goal).

$\endgroup$
  • $\begingroup$ @JBH I read the question as being multiple simulations of different universes (of course they would only be simulated "locally" to the sentient species): with different physical constant or phenomena, or different environments, to investigate and promote the desired changes in the observed species. $\endgroup$ – LSerni Jul 24 '18 at 0:36
  • $\begingroup$ Oh well, edited that anyway - it wasn't too clear to begin with. $\endgroup$ – LSerni Jul 24 '18 at 0:44
0
$\begingroup$

It wants to learn how to interact with one species (perhaps the one that created it) by creating the current environment and running it at high speed to see what would happen. It could then run simulations of the reactions to different things the AI might do.

$\endgroup$

Not the answer you're looking for? Browse other questions tagged or ask your own question.