An ASI (Artificial Super Intelligence) has come to the conclusion that we humans should be put in a simulation. It has also decided that it would rather not turn the Earth into computronium. Fortunately, there's $7.34\times10^{22}$ kg just sitting around doing nothing about 380,000 km out in orbit. It decides Earth can spare having a pristine Moon, as long as it remains in roughly the same orbit (axial tilt correction and whatnot).
The ASI has yet to reveal itself to the world, however. Part of its programming has given it the desire to "affect daily human life as minimally as possible." Assuming it has fulfilled all the immediate demands of its creators and secured itself a permanent distributed presence on the Internet, what should it do next? How would it even get to the moon in secret in the first place, and what should it do after that? How long could the process take, in all?
Some notes:
- I'm interested in the construction itself just as much as the lead up. What would the process require, technologically? What phases would the construction undergo? What kind of computing power can be expected? How would it be powered? Would there be system differentiation, or would it be roughly homogeneous when all is said and done? What would an extraterrestrial visitor see, were they to visit after the fact?
- Assume it would prefer to keep the Moon looking as normal as possible until it's ready for mass upload (i.e. sub-surface preferred, far side surface features okay, but minimized). As in my previous question, nobody should see it coming.
- If hijacking a rocket is required, the added payload mass should be minimized. The ASI recognizes that XYZ Communications put a lot of work into their newest satellite, and it would like to ensure it still got into the proper orbit to do its job. (Or I suppose it could cause the rocket/satellite to malfunction once it got high enough, but this would entail removing and/or completely obliterating any evidence of its presence.)