I know they've said as much, but it's debateable to what extent their statements can be taken seriously. They certainly haven't scanned every brain on the planet, so they can't have produced a completely accurate ai copy of every real person.
That's beside the point. I'm not talking about whether they're exact copies of individuals, but about whether they're sentient beings in their own right. That's purely a matter of whether their brains are equal in complexity and structure to human brains in general, regardless of whether they match specific individuals' brains. After all, two different humans' brains are not exactly identical to each other, but that has no bearing on their respective sentience, because both their brains are complex enough to be sentient. So these may be new people who just approximate the original ones, but they're still people.
There's also the question of what 'accurate simulation' necessarily means. It could be there are individual ais standing in for every person. It could be there are ais standing in for some people and algorithms being run by a massive program to 'simulate' other people and areas that aren't currently required to be fully focused on.
But that's not simulation in the sense I understand it. Scientifically speaking, a simulation is a model intended to accurately replicate a real-world process, to predict how that process would actually unfold given a certain set of starting conditions. For example, in making computer-animated movies, they hand-animate things like character movements and expressions, but they use mathematical simulations to generate things too complex to animate, like flowing water or the behavior of hair and fabric in response to gravity, motion, wind, etc. They enter a set of starting conditions and let the simulation generate a result on its own, and if they don't like it, then they enter new starting conditions and physical parameters and run it again.
So I don't take "accurate simulation" to mean an exact copy; I take it to mean a simulation equal in complexity and physical behavior to the real world, so that it behaves in the same way reality would behave given the entered set of initial conditions. What's accurate is its prediction of the outcomes of a given set of conditions, its modeling of the way the real world would behave.
It could be there are no individual ais at all, and every person and action is being simulated by one massive program which is somehow able to perfectly mimic millions of different personalities simultaneously.
That's a spurious distinction. All that matters in defining a mind is its structure and processes. It doesn't matter what it's made of or whether it's a part of something bigger. If the single massive program is accurately simulating the activity and structure of millions of distinct sentient minds, then those are distinct sentient minds for all intents and purposes.
There are scientists who argue that our entire universe is statistically more likely to be a simulation than a physically real universe. The idea is that a real universe could contain countless computers advanced enough to simulate entire universes, so the number of simulated universes would vastly outweigh the number of real ones. The Framework just needs to simulate the Earth, which is comparatively easy.
I suppose, though I find it hard to imagine how that kind of programming could be hard-wired.
That was just one possibility. The larger idea is that the workings of the brain are not one homogeneous mass, but a bunch of different processes working together. And that means that you can transfer some without the others. For instance, there are several different kinds of memory -- there's episodic memory (recall of specific events and personal experience), semantic memory (recall of factual knowledge and the meaning of things), and procedural memory (recall of learned behaviors and skills). Which is how someone (at least in fiction) can have amnesia about their life experience and personal memories but still remember how to speak English or drive a car. Similarly, there's a difference between our conscious decision-making processes and our more primal drives and reflexes, and a separation between different aspects of decision-making. For instance, I read a while back about a form of brain damage that could impair one's ability to make moral decisions -- when presented with a problem that involved deciding whether to let someone die (in the abstract), subjects would lose their normal reluctance to make that choice. Sort of like that
Voyager episode where the bad guy turned off the Doctor's ethical subroutines and turned him into a psychopath.
So it isn't necessarily a given that transferring Aida's mind would require transferring the whole thing without modification. Some bits could be left behind.
