THE A.I. TAKEOVER CONSPIRACY
My Gripe: Elon Musk, Bill Gates, Ray Kurzweil, all these luminaries
waxing about how AI will take over and it'll be lights out for
I mean its enough to chap one's hide...
Their field of misinformation and slop-thought is so -rich- (sorry,
couldn't resist that)...where to start...
How about how Gates (29 Jan. 2015) reports that he can't understand
why other people are not worried about this.
Here's one physicists take on the confusion. I will say it right
here right now: Machines will never take over and kill all the
humans unless we humans want them and make them too do so, abetting
them with our active, continuous support. There is no convergence.
If you want to be scared, be scared about something real that we
know about: namely large impactors.
AI Takeover? Yeah, I'm not too worried and here's a few reasons:
1) (What's the real deal:) Machines are reflex engines only. Human
brains are not just reflex engines. Notice I didn't say what human
2) (Technical reason 1: Machines waaaay too primitive) Reflex
engines are printed (basically really still 2-d) and large. Biology
is 3-d and basically works across all scales from atomic up. It is
not great, but there is nothing like it that we've ever made.
period. Manufacturing is still in the stone age.
3) (Technical reason 2: Category faults)
Information is not knowledge
Knowledge is not understanding
Understanding is not Wisdom
point is, reflex engines are really at the first level in the
heirarchy. Now, of course, they do not have to master the heirarchy
to kill us all. True. Problem is, since we can ascend the heirarchy
we can probably get them first, as we may have a broader context to
more early identify the problems and make a change.
4) (Technical reason 3: Frailty in deep time) Humans cheap, feed on
grasses mostly, each Human/grass a replication bomb. They all come
from deep time.
Reflex engines need a big wheel turning somewhere to power them. Not
clear if you can get through deep time without something like
sex...but maybe they don't have to; just kill all humans and
reach a stasis?
5) (Technical reason 4: Ill-equiped at the physical layer) One gets
the sense that these folks who purport to believe in the great AI
takeover are enamored with how facile computers are at changing
logic levels, again, the reflex engine layer. But there are real,
significant problems at the physical layer, as one immediately
recognizes if you try to build a robot. The motors and hydraulics
and pneumatic actuation (ideas that are now hundreds of years old)
we use in robots is generally of very limited capability/utility
compared space for space and power for power with actuation in a
muscle "fabric". A person can do a somersault, turn around and
thread a needle. Quick. Note these items area doing work at
significantly different force/motion scales. Machines? My point is that
there is a basic
mismatch between how we now actuate robots and the needs in the
world around them. Hey, its only been 150 years since the electric
motor...maybe improvements coming? Not! Probably a pretty mature
6) (Technical reason 5: Reflex engines are soon to be peaking
7) (Technical reason 6: