Tuesday, September 18, 2012

The simulation argument meets NASA

THE SIMULATION ARGUMENT is being defended by NASA scientist Rich Terrile, who is interviewed by Ben Makuch at Vice: Whoa, Dude, Are We Inside a Computer Right Now? This NASA Scientist Thinks We Could Be. Excerpt:
VICE: When did you first surmise that our reality could be a computer simulation?
Rich Terrile:
Unless you believe there’s something magical about consciousness—and I don’t, I believe it’s the product of a very sophisticated architecture within the human brain—then you have to assume that at some point it can be simulated by a computer, or in other words, replicated. There are two ways one might accomplish an artificial human brain in the future. One of them is to reverse-engineer it, but I think it would be far easier to evolve a circuit or architecture that could become conscious. Perhaps in the next ten to 30 years we’ll be able to incorporate artificial consciousness into our machines.
These are interesting speculations, but they are just speculations. Note in particular, the first clause, "Unless you believe there’s something magical about consciousness," which is a roundabout way of saying you don't understand what consciousness is and neither do I, therefore I'm going to run with my assumptions. Fair enough, but assumptions they remain. We have no theory of consciousness that has any explanatory value or that could lead to the ability to replicate it artificially. To say that it is an emergent property of sufficiently complex computation is to commit the fallacy of obscurum per obscurius, explaining one thing we don't understand (consciousness) by something else we don't understand (emergent properties) and it simply sidesteps the real problem of why some hunks of matter (or at least the one I'm in) acquire a sense of "me-ness."

Moreover, complex computation and even mentation are not by any means necessarily connected to consciousness. Anyone who has experimented with Raja Yoga (and, really, who hasn't?) has learned that most of the thought processes running in their own mind are unconscious and are only briefly illuminated from time to time by the narrow sliver of consciousness that inhabits their mind and flits about it to use the mental processes as needed.

It may be that (1) consciousness is indeed the central processing unit of the physical brain—but we can't show yet why the CPU has its own sense of existence distinct from the mind. It may also be, as the ancient yogis thought, (2) that consciousness (which they called the purusha) is something external to the material world and cannot be explained in material terms, in which case we will not find an explanation of it. Lots of metaphysicians have problems with the idea of immaterial entities, but I'm enough of a Platonist that they don't bother me too much. It is also possible that (3) there is an explanation of consciousness in material terms, but because we process everything through our consciousness, we are in principle unable to understand the explanation any more than we can look in our ear with our eye. If 2 or 3 apply, we won't be populating any simulated worlds with conscious beings.

We not only do not understand consciousness, we also don't have any way to measure it. The Turing Test is a side issue. At best it would measure the ability to think, which is not the same as consciousness. Also, in practice, even today some chatbots can fool people into thinking they (the chatbots) are human.

So the case remains to be made that increases in computational power will eventually pass some threshold that creates conscious AI and allows us to simulate worlds with conscious creatures in them. But as, if memory serves, Nick Bostrom pointed out years ago, that doesn't preclude the simulation argument. The universe may indeed be a simulation with a limited number of player characters inhabiting it (me at least, plus you if you are reading this), but with everyone else as non-player characters who are simply realistic simulations with no internal consciousness.

I return to the simulation argument from time to time, because similar ideas are at the basis of ancient Gnosticism (the kind associated with the demiurgic myth): the universe is the flawed creation of a fallen creator and we are Pleromatic (heavenly, more or less) sparks of light that have become trapped in the material muck. The object of the game is to gain the knowledge (gnosis) that will allow us to escape from the fallen created world and return to the Pleroma.

Back to the article, for some final theological reflections from Terrile:
You seem really at peace with this concept. When I first heard about your theory I was incredibly bummed but, obviously, intrigued.
I find great inspiration in it, and I’ll tell you why: It tells me that we’re at the threshold of being able to create a universe—a simulation—and that we in turn could be living inside a simulation, which could be in turn yet another simulation. And our simulated beings could also create simulations. What I find intriguing is, if there is a creator, and there will be a creator in the future and it will be us, this also means if there’s a creator for our world, here, it’s also us. This means we are both God and servants of God, and that we made it all. What I find inspiring is that, even if we are in a simulation or many orders of magnitude down in levels of simulation, somewhere along the line something escaped the primordial ooze to become us and to result in simulations that made us. And that’s cool.
That's a much more positive take on Gnosticism, with demiurges all the way down, so to speak. Who knows, maybe the Gnostics or (and?) the simulationists are right. Call me when you have a real theory of consciousness.

Background on the simulation argument and Gnosticism is here with many links.