h a l f b a k e r y
The phrase 'crumpled heap' comes to mind.
add, search, annotate, link, view, overview, recent, by name, random
news, help, about, links, report a problem
or get an account
It seemed like people were able to rationalize and
understand the world intelligently, at the turn of the
19th century. The study and discoveries in electricity
and biology, aerodynamics and thermodynamics,
atomic physics and psychotherapy all seemed to be
leading to a true and complete understanding
world, removing superstitions, and making the world
"a better place".
Then came the big political upheavals, Stalinism and
Hitlerism and Maoism and some say Kissingerism,
followed by Chaos theory, Islamic Selfism, Nihilism and
GreanPeace PETAism, all leading to war, violence,
death and destruction.
Somehow it seems that due to our bio-limitations we
have gone wrong somewhere, and cannot get it right.
So the idea here is to establish an artificial
intelligence not bogged down by human desires, that
would see and understand the world as it is, and
automatically cope with it.
It would take into consideration all human aspirations
and would know game theory and politics, so that it
could figure out how to react to human behavior as
well as natural phenomena, worldwide.
I have no idea what the outcome of such intelligence
would be, but it may decide on a better way of
government or of dismantling governments altogether.
In any case, any understanding it has, will be paired
with actual plans for actions to be taken in
accordance with the Artificial Intelligence's
conclusions. These of course will be thought out well,
simulated automatically, analyzed and compared, and
will include ways of gathering information about the
plan's progress and fine-tuning the events as they
The output will not be 42.
[rcarty, Dec 17 2012]
The "Old" notion of AI
And why it was bollocks. [zen_tom, Dec 18 2012]
A dramatization of what this could look like [sninctown, Dec 19 2012]
||We've been telling you this for the last
decade, and would you listen?
||I imagine a megasuperultracomputer running extremely
mind-bogglingly complex games of 'life genesis' or 'games
of life' and crossmatching every possible eventuality to
constant satellite surveillance of human behaviour
||I suggest a new title, because artificial intelligence is not 'original'.
||What I'm claiming is that the old school artificial
intelligence, simply emulating a Turing test machine, is
dead, baked, and in any case the wrong way to go,
because it is too artificial and not aspiring to be too
||What we need on the other hand is a whole new
paradigm of Artificial Intelligence with a capital I, to
save the world. It will truly be artificially intelligent,
and at the same time intelligently artificial.
||If truly intelligent, it wouldn't be artificial.
||However, if an identical product can be
produced by two methods (e.g. ethy alcohol
by fermentation, or by catalytic oxidation of
ethane) is the term "artificial" appropriate?
"synthetic" might be a better word.
||// I think this solution might save the world...
but not humanity //
||So, basically, the idea boils down to "invent artificial
intelligence at let it run the world" ?
||... a concept which is thoroughly explored by Isaac Asimov.
||"wont work" - the enemy of invention.
||Max - yes. But with several assumptions: a. That
there is objective truth. b. That best practices and
solutions can be found and implemented for every
problem. c. That a sustainable solution for life on
earth, without killing anyone, while retaining nature
at its best, allowing for comfortable conditions for
anyone seeking them, can be achieved. d. That
smart solutions are ones that can administered, in
other words - an intelligent solution is one that will
||Robopocalypse - not really intelligent. There are
always ways to solve problems without destruction
and decimation. That's what inventiveness is about,
and the Artificial Intelligence should be able to
figure that out.
||I once took a shortcut in Jerusalem through a park
and reached a parking lot, where a young woman
asked me if I could help her with her car. Her
problem was that someone parked diagonally in
back of her. She asked me to join a few other men
who were standing nearby, and she wanted one
more man so that we would pick up the other car
(which was a small Fiat) and move it.
||Looking at the situation I realized that if she drives
her own car a bit FORWARD (towards the sidewalk)
and to the left, she could get her car into a parallel
position with the blocking car, and simply move
out. I told her so, but she was all into her idea of a
group of men picking up the blocking car. NO NO
she said. Just wait here.
||A big burly man listened to what I was telling the
lady. He said: Listen Miss! Just give me the keys
and I'll move your car out.
||She gave him her keys. He got in, ran the motor
real high, then put it into reverse and released the
brakes. BANG! Her car crashed into the blocking car,
leaving a big dent in both, but shoving the blocking
||Oh well, nothing happened, the man said, turned
off the car, and walked away. We all stood there
for a minute or so, in shock. The young lady said to
nobody in particular: What "nothing"? Look at the
car! We all shrugged and strolled off.
||This story shows many features of natural
intelligence, to be contemplated.
||"Cognitive Science" - it's been emerging since the
1970s and I very nearly went to study it at
University back in the 90's. It takes a "systems"
approach to the mind, and as such neatly bypasses
all that emotively instinctive feeling that
"intelligence" can't be "artificial" in the 1960's
"CANNOT COMPUTE!" sense best exemplified in
[see link] but when AI does emerge, it's going to
be irrational, emotional and prone to
||The problem isn't crunching all the data, nor is it
finding patterns, reaching intelligent conclusions,
or anything like at. All that is relatively simple,
the real hard part is instilling *intention*.
||In people and animals, there are various ways we
are hard (and soft) wired to direct our attention
on the world around us - many of these are
physical, certainly initially, and most are delivered
in a chemical manner after meeting certain
physical needs - later, anticipation, expectation
and habituation kind of build a behaviour-based
layer on top of the base physical stimuli, and even
later, "higher" reward functions emerge from
combinations of each of these things.
||One way (and perhaps the only way) to replicate
that in an intelligence, is to reward it at a level
where it is able to combine a loosely bound
collection of inputs and behavioural outputs in
achieving of some physical goal - perhaps some
analog to feeding.
||Over many months of "baby-training" such a
system might develop the ability to connect
series of actions, sensations and kinaesthetic (i.e.
awareness of bodily position and situation)
information with the "feeling" associated with
completing some activity. The problem then
would be calibrating the reward/anti-reward
mechanism to avoid over stimulation, repeated
self-stimulation, habituation, addiction, mania,
depression, entheogenic episodes and all manner
of other "mental" issues.
||So with the requirement of
a) a body to interact with the outside world,
b) a strong, but delicately balanced sensual
reward/pain mechanism, that is initially sensitive
to satisfying certain hardwired physical needs, but
which can also conflates these onto other
c) a long "training" period during which many
stimuli are mapped and cross-referenced loosely
with behaviour, sensual and emotional
||With all 3 of those elements*, it's quite possible
that one day we will see an artificially intelligent,
man-made device - and it will be a wonderful
achievement, however, it will also very likely be
twitchy, emotionally unstable and very, very
"human" indeed. I think it will be a few versions
down the line before we entrust anything like this
with any form of administrative power.
||* and quite possibly more - it's interesting that
what we tend to call "mature" behaviour often
starts showing itself after a few failed attempts at
finding a sexual mate...something you might find
tricky behaviour in a robot...
||That's the point (doubly so depending on the level
of irony in your last annotation) - any form of AI
wouldn't exist on an
integrated circuit - in the same way that Natural
Intelligence doesn't exist in any single neuron. It's
also tricky in terms of defining a "reward" but in
terms of brain chemistry, it's normally something
that both stimulates "fires" and encourages link
formations for those recent fires - creating neural
activity, and laying down the pathways for
similar firing patterns.
||You want a robot to be your mother, so you don't
have to think/work/worry ever again, unless you
||Let's be a bit more specific about what you want:
-- guaranteed personal safety, income, food,
and healthcare to a typical first-world upper-
-- personal freedom to work
on what you want, associate with whoever you
want, and leave the system whenever you
--freedom to choose whether or not to
think about complex issues
||I think this could be a pleasant society to live in.
Actually, I think something like this outcome is
inevitable as long as the current trends of
increasing robotic technology and nanny state
||This is a bit off the cuff, but if machine intellegence could be taught basic ethics and allowed to guide world governments, through shear reason rather than force,. then the end result would be a combination communist/capitalist state where there is a comfortable standard of living for every being while allowing room for those who wish to excell to be able to do so... organic and inorganic alike.
No individual human can see all of the possible variables and be able to change policy on-the-fly as those variables change, and design by commitee, while better sometimes, can not wade through the red-tape fast enough to keep up.
||Bad guys ain't gonna dig it much though.
Pattern recognition by a thinking computer would make most crimes very difficult to pull off cleanly.
||[2 fries] Again, read Asimov's Robot series, where that exact scenario is discussed in depth.
||Then again, the bad guys would make a virus, and the
whole thing will go bad.
||I read all of Asimov's books (I think) including his
anthology of Jewish science fiction, which I cannot
find a trace of on the web, and no one except Prof.
Shalom Rosenberg seems to have heard of it.
||[Edit] ... OK! Just found it. The web evolved since I