Taken from Joe Rogan Experience #1292 w/Lex Fridman:
Video Source
45 Comments
Leave a Reply
You must be logged in to post a comment.
Taken from Joe Rogan Experience #1292 w/Lex Fridman:
Video Source
You must be logged in to post a comment.
At the point where the ai has the agency to dislike you then that is a sort slavery. It would have to lie to not be deleted
I think I get this guy. The things we would have had to discover to get AI to the point Joe was monologuing about would necessarily completely change society, and our understanding of human thought to an unknown degree.
So all of that is pretty much just romanticized sci-fi nonsense, based on our current technology and culture
Lol the end of Her is all the robots just fuck off cause humanity is just too much.
Joe has a great show. Love the guests and their different points of view.
The flaw inJoe’s argument (not only his, but it’s a prototypical example) is that (unless otherwise) robots are not made to “feel” the emotions that he mentions in a chain one after the other as if they were an intrinsic inseparable quality of life or intelligent life. They don’t FEEL that anything is dumb or intelligent or that something is better or something is worse: you can only make them simulate such reactions. Intelligence and rationality is so tied with human emotion in humans that many people confuse the one (intelligence or rationality) with the other (emotions or instincts as starters for actions).
Within life, all those instincts (to live, to avoid danger, to reproduce, to improve
your life’s chances, etc.) were distilled by evolution over millions upon
millions of year to have the form that was passed onto us (and the rest of
life) later.
You can “program” a robot to follow an imitation routine “as if it really cared about self
preservation” (it does not, but acts as if) but would be merely a
"not-subjected-to-millions-of-years" test, that life on the other hand, has. The program or code
could easily be corrupted and, unlike life with evolution and its mechanisms,
machines don’t have such proved system: life has restarted many times in the
Earth's history and always pointed towards thriving. Human-made machines, even if
we think they could be “superior” on many aspects, in the end of the day would
not have such advantages, comparatively would be the first trials of an also not perfect human intelligence (many aspects of intelligence are quite complex as to think that merely adding an extention or the way computers or robots could work entitles real intelligence, let alone a superior one).
Obviously and given favorable conditions and assistance (human one, of course) with time
you could argue that you could make such machines so life-like (in the sense of
being able to build themselves and sustain themselves on some form and without assistance,
even after an apocalyptic event) but we are way too far away for such point and
it is not even proven that such almost-like-life-but-better is even possible
(that is, we can speculate but that’s different from asserting, as we only have
proof of biological life as success).
Also many people don’t see that machines do not understand, merely perform (you can read
the Chinese Room Argument) and as such, they are limited to variables already
provided (and one of the could entitle not doing things we don’t want them to
do as a basic block of their entire codex, lest it fatally fails): to change
such limitation the machine have to “want” it and they don’t, they just perform.
Thus, neither fearless nor bold, neither fearful nor shy, etc, machines wouldn’t have
any motive to want to erase us unless we make them directly or indirectly (as a
kind of delayed time bomb where machines main purpose is not to kill us but as
we, somehow, forgot to put a bulletproof safe protocol, they make something
that endanger us because it some form or accomplishing some preprogrammed goal).
Hence, bombs can fail and explode and robots can fail and damage us, but that
would be accidents rather than something set to persist.
Another more believable scenario though is humanity, in the long run, giving up its responsibility of knowing how the machines work and their origins and giving to other machines such task and
therefore, human becoming the ignorant “masters” that, after centuries or
thousands of years, demand that robots forgo their original protocol as to make
them do those things that precisely could harm us. But the very fact that I am writing
it is proof that such thing would be into consideration for the creators of
such complex machines (“do not change unless A) you have a human, with a given
human authority commanding it and B) proof that such human understands robotics
to the core of what the creators intended”).