September 10, 2010

Why Petraeus Spoke - The Daily Dish | By Andrew Sullivan

September 2, 2010

Jaron Lanier Inspires This Takedown of AI, via Modal Realism


There is a glaring problem with the idea that all human mental processes (not just thinking as such) can be modeled in terms of computational intelligence, etc., like computer programs.  Computations work with numbers in effect, and can't represent anything that isn't some sort of mathematical structure. Well, as modal realists have explained: the very idea of substantial existence ("real stuff" being distinct from Platonic forms) cannot be coherently defined and explained in strictly logical terms. We not only can't explain why some possible worlds, have a special trait called "really existing" and others don't, we can't even explain what the distinction consists of to start with. Indeed, we speak of the "existence" of defined configurations of numbers and shapes. Hence thinkers like David Lewis said all such worlds "exist" with equal standing.

This leads to the IMHO hilarious irony for "materialist" (?) thinkers like Dan Dennett: if our minds are only computational, then we can't even conceive a "material universe"! (Part of Dennett's assault on ideas like qualia and phenomenal sense-data is to tease their ostensible realness by saying they're made of "figment."  Well, as a tweak to Dennett I will  use the term "mysterial" for the alleged special property or "stuff" that imbues mathematical models in the MUH ultimate ensemble and makes them "really exist" - like ours is intuited as - instead of just being "models" like for if I wonder what physics is like if space had 23 dimensions.

Note that the thinking done in a model simulation is just like that in a "real world" - if both are defined in AI terms.  Hence, the AI entity can't even frame the thought, "I am a material being and not just part of a program simulation or  Platonic structure of math (as in MUH.) It is ironic, that those who believe in AI concepts of human thought *have to be modal realists* to be honest and can't even be genuine "materialists." They implicitly don't believe in a way to conceive of substantial worlds being distinct from the mathematical process itself (although most don't realize it and continue to pretend they can make the distincion.)  Hence, all such worlds exist and there is no point in "physics" or materialism per se.  Even all cartoons would be real worlds.

Jaron Lanier, philosopher of mind and cybernetics, made a similar point (awaiting cite here).  His was more on the order of, a mind wouldn't need to be embodied since the Platonic representation would be equivalent.  If so, then all those minds have the same thoughts and experiences we do just "as the same math."  It is sort of like the brain in a vat, except this time the question is: are my thoughts happening in a "real material brain" (at all, anywhere!) or am I just the pure thoughts themselves?  I take this to the next level: not only would the mind not know or be able to find out whether it was a material manifestation or not, it could not even have the intelligible thought of what the difference meant or consisted of!  I suggest checking Lanier's new book You Are Not a Gadget.

Note that we have Bayesian expectation problems about own in such a case: we'd be more likely to be in a world that  was just coherent enough to create us, but no better (and not inclined to stay that way - since changes are of course "describable" features of a concept world too!) See below re Marcello Gleiser.

Neither can we logically define "real flowing time" as a special way for reality to be, distinct from simply there being a configuration of points and lines (world-lines) in a "space" of four-dimensions. Yes, we *represent* time events in math, but there is no definition of it as a distinct entity, no mathematical way to point to the "t" in dx/dt and say, "That is some special, qualitatively different trait and not just another axis in a coordinate system; in kind like the "x". Yet somehow our minds grasp or profess these distinctions, as they do the idea of "consciousness" being special. I don't think that's a coincidence.

And if a CI/AI proponent is going to gripe that our minds are computational and that "consciousness" can't be logically defined or explained, then neither can "real flowing time" or even the idea that for example our thoughts happen in "real material brains" and not just some Platonic math structure - as I explained in amplification of Lanier's point.  I call this dilemma "the brain as a fact" (as mere computations  distinct from "incarnation") in comparison to the old conundrum, how does any of us know he or she isn't really a brain in a vat being fed simulations Matrix-style.

Now, this argument doesn't prove that either AI/CI or MR/MUH is wrong, it just ties them together in ways that may not be palatable for e.g. conventional materialists who believe in AI (Dennett?)  My final word on this, as a non-believer in either of those perspectives:
I feel, therefore I (really) am!

PS if you have access to Facebook, check out the discussion at William Hodge's thread.  It's based on this
web article by Antonio Damasio at bigthink.com

I'm working on some format issues and may twiddle this post as per my custom. 

Labels: , , , ,