Slashdot Mirror
The Emerging-Behavior Debate
weezer writes "Interesting article about 'emergent behavior' of complex robots/devices. This was all over the news tonight too: theories about machines being able to "Think" on their own. Read more about it. " What do you folks think (no pun intended) - how much differently will machines/devices think? Or do they?
Why People Think Computers Can't
We've known this about AI for some time now. You can find examples of 'emergent behavior' in simple games, like Conway's Game of Life.
:) but whenever I do it by hand, it seems much harder...
From a few simple rules and their interactions, much complexity springs. It only takes maybe 12 rules to do predicate calculus, (not much PROLOG at all
The problem is that not everything can be neatly quantized into rules. That's the problem that the Cyc project has always faced, and they're probably closer to getting a self-learning system than anyone is.
I would love to see AI produce results, I'm sure machines could think and solve complex problems, reason and reach conclusions as we do, but just encoding the basic knowledge we take for granted is a huge task, and trying to make a machine with the capability of doing all of that itself is an even larger one.
pb Reply or e-mail; don't vaguely moderate.
Just because something doesn't display HUMAN qualities doesn't mean its not INTELLIGENT!
(Why am I writing in CAPS?!)
Humans have a very narrow understanding of intelligence. Our basic inferiority complex puts us on the highest scale of intelligence, and yet we can't even communicate with dolphins.
When computers are intelligent, it's going to be very hard for the majority of people to be able to deal with it. There are many reasons for this.
* Anything that removes the concepts of free will and human speciality destroys the basis of many religious and philosophical belief systems.
* The emergence of an intelligence which is non-human based is threatening to our self-centered view of the universe.
* Acceptance of computer intelligence (which is modular and hence expandible) puts the limits of human intelligence right in our face.
Computer intelligence is coming. I don't think we are ready to deal with it yet. It wont be like talking to another person... although we will try to make it that way.
neo
ps (If anyone knows how I can get in touch with those damn matrix guys... let me know. I want my name back.)
Why is it that every time people start talking about AI, you get a bunch of closet philosophers who proclaim that, "If we do create a truly Intellegent machine, it will invalidate all the world's religions." I just don't get the connection. How is it that the ability to create a machine so complex that it becomes aware of its own existance will suddenly change the fact that there is or isn't a God (or gods) and people's fundamental measures of right and wrong? I'd say that these people have a very narrow view of what religion is.
There will always be things in this universe that are beyond our comprehension, and as long as this is true, people will struggle to explain what cannot be explained. Some will turn to science and the ability to reduce everything to rules and formulas. Others will turn to religion and the belief that there is more to this universe than we can see and touch and measure. The existance of "thinking" machines will certainly complicate this search for Truth, but it will not make one side or the other go away.
Your Servant, B. Baggins
Like complexity, emergence is one of those topics that no one can agree on how to define. If you ask a holist, s/he will tell you that an emergent phenomenon has top-level behavior that cannot be predicted from the bottom-level description. However, this definition seems to discount a simulation as a method to predict that unusual top-level behavior can arise.
;-)
If you ask a reductionist, s/he will tell you that nothing is emergent. In fact, Marvin Minsky's Society of Mind is often cited as a model that explains how intelligence can emerge from dumb lower-level building blocks. However, Minsky is a firm reductionist and claims that nothing is in fact emergent. (Ask him yourself.)
In any event, this article would seem to imply that emergence is a recent discovery that has placed us within an epsilon of building an AI. Like most mainstream articles on science, it's very much out of date. People have been actively talking about emergence for decades. And while we have gained a great deal of insight over that time, we still do not understand how the visual cortex works (which is the best-understood portion of the human brain) let alone more complicated things such as language and planning.
But none of this means that building an AI breaks any of the rules. Evolution took a few billion years to go from a single cell to multicellular beasts, and another billion years to produce a beast capable of talking about itself. Computer Science has only been around for a few decades, so give it a bit more time.
(All of the above was written by Gary Flake's personal agent
The Computational Beauty of Nature
The word "System Administration" will take on a whole new meaning....
NT Server: You never listen to me! All you do is play quake and read that nerds for news site! How hard would it be to just E-MAIL me once and awhile, huh?
3com switch: Look buddy, you keep me here, locked up in this closet all day.. I get no respect, and all I here is "route route route, all night long, route route route, while I sing this song"...
Linux Server: Now, see here, check out all these graphs of my performance. I just finished that cronjob, balanced out the www sever load, and started the espresso machine. All this BEFORE lunch. Debian 6.3 rules!
Windows 2010: "I can't get no.. do de do, satis-faction, do de do"
Cisco Router: I take the packets in, and put the packets out, and I shake 'em all about.. I do the packet-pokey and I ACL alot, that's what it's all about....
Sun: Does this add-in card make my butt look big?
--
You can't write a program and expect artificial life to spring forth from it. A natural brain doesnt use mathematics and predictable systems to process. All your clusters of neurons are just one on/off switch. Anything complex is distributed over the system and solved in mass numbers instead of one centralized system. Sure these systems eventually become non centralized, but they need to start off as separate units and then cohese (is that a word?) by themselves. You have to teach them that ramming into walls is bad, not program them not to run into them. Thats real emergant behavior. ntelligence is writing your own code, not follwing prewritten code. To create artificial life we need to start where natural life on this planet started, back in the primordial ooze. It took 4.5 billion years of code to get where we are now, a hop skip and a jump away from figuring out not to run into walls. With AI we need to start smaller than small and work our way up. Robodyne Cybernetics thinks they have more than they do.
I'm a loner Dottie, a Rebel.
Here's something to chew on, the universe is irrelevant. Any quantifying of any portion of the universe limits the mass view of the universe. Saying "there is a god" or "there is no god" gives you only a keyhole view of everything, it limits your perceptions down to the point where everything has to be qualified according to your particular set of beliefs. Arguing that you're self aware gives you the perception that you are self aware and anything you perceive becomes the product of a self aware mind. Believing you are not self aware makes everything you perceive go through that same little filter just with different parameters. Never say something is impossible until it actually becomes impossible and even then hold out saying it's impossible because there is always the inevitability that it is possible. If you dont understand what I said, then dont read it. I dont want people flaming me saying i'm wrong and you're right simply because your perception of the universe means about as much to me as that glass of tea I just drank.
I'm a loner Dottie, a Rebel.
Given the definition of :
;-)
"Today, "emergent behavior" is often used to describe computer systems grown so complex they exhibit capabilities not programmed in."
Then the question is : are bugs of windows programmed in or not ? If not, then does my Windows box think ?
Maybe Windows 2000 will be Space Odyssey's HAL... oops ("I can not let you do that David !"
The world belongs to those who get up early. - I'm far from being the king of Earth then
read Isaac Asimov "Robots" novels
:-)
(and by the way also read the "Foundation" sequel...
The world belongs to those who get up early. - I'm far from being the king of Earth then
Before we go looking for artificial intelligence, why don't we prove the existence of natural intelligence? I'm not sure I've ever seen this done. Every real argument for the existence of mentation and identity I have ever seen (admittedly this is limited to a few undergraduate philosophy courses) from "The ego posits itself" to "I think, therefore I am" are basically tautological.
I'm not saying AI research is wasted effort, but I think its greatest value lies in how it helps us try to figure out the nature of our own intelligence (if we have any...)
Considering I have a degree in Cognitive Science and Artificial Intelligence I think I know what I am talking about here. First, there is NO program where the behavior is unexplainable from the programming, you simply have to see how the software _reacted_ to the environment. It might do something that even the programmer wasn't expecting (this happens often) but this can be explained by analysis of system logs etc. and all follows rather simple rules. Emergent behavior _seems_ like a really neat explanation of intelligence and thought, but it really just moves the question: where does it EMERGE from, and how? There are NO explanations of emergence that offer anything other than interesting analogies, e.g. a car is an emergent property of all the underlying parts - i.e. no one part is a fast moving vehicle. Okay that seems logical, but the difference is that the whole, while greater than the sum of the parts, is explained by them. This is not the case with the mind. No one knows how a conscious being arises from unconscious matter, and there are very good arguments that we may NEVER know. Not because we are not intelligent _enough_, but simple because we have intelligence in the first place.
-Jeff Rankine aka ShieldWolf
My $.02
just = (My)Opinion.toCents();
The article looked good until the first 'proof' point - Robodyne Systems, and their 'fractal robotics'. Puh-lease, I wish jernelists would take the effort to research their stories.
Joe Michaels is the laughing-stock of comp.robotics.misc, his fractal robotics ideas are whoppers, but the technology doesn't exist to perform what he touts as simple coordinated movements. He once had a movie on his webpage detailing how two box modules could move around each other - the very cornerstone of what he proposes with fractal robotics. Unfortunately the top box was being pulled about with an almost undetectable string.
Joe's rants and raves in the c.r.m newsgroup have alienated him from the robotics community. There is real research being performed in the areas of genetic algorithms, emergant behaviours, and neural/fuzzy logic systems - but there's nothing close to machines achieving self-realization.
ttyl,
Why must an AI pass any Turing test to be considered intelligent? Given that there are an infinite number of states in the universe (or at least enough that we could not comprehend understanding all of them entirely), why would we ever expect a machine to be able to interpret and respond to every one before we can call it intelligent? Humans do not do this. We are not pure algorithmic processors. Humans, and other biological intelligence use highly complex fuzzy logic...we use neural networks...we adapt to our reality as necessary...we are not required to understand it in entirety before we can live in it. This is where intelligence "emerges"...we are intelligent because our fuzzy logic leads us to conclusions that we would not otherwise be able to come to via purely algorithmic processing (we simply don't have the hardware!)
So one of the current theories (not really current...it's been around a while) is that by making a computer or robot with processor(s) that reach a level of complexity roughly equal to the human brain, then intelligence will emerge. Anyone else notice how fast processors are gaining in speed and complexity lately? If we continue improving them at the current rate, then in another 5 or 10 years we'll probably have a true artificial intelligence.
This raises some questions....
Does it have the same rights we do? Can we control it? Do we want to? Should we allow AI's to exist?
Think about the advantages they'd have. The human mind is an extremely powerful tool, but it has it's limitations. We don't multitask very well. We can't do basic math at mind-boggling speeds. The list goes on and on.
An AI of approximately equal intelligence to the average Joe would be able to outthink the greatest of human minds. We'd be unequipped to defend ourselves against it.
Food for thought.
-Andy Martin
-Andy Martin
If y'all don't like me, blow me.
Lets get some terminology right here.
According to Mr Flake wholists will tell you that emergent properties cannot be predicted from the botton level description and reductionist will tell you that nothing is emergent and that everything can be explained from dumb low level building blocks. Well and good. But this implies that these two terms of opposition are the only two alternatives. It is possible to say that higher level properties are emergent and explainable in terms of their lower level properties. The middle position is that higher level properties supervene on the lower level ones. Supervenience allows for the possibility ontologically robust higher level entities while allowing for these entities to be explained in terms of more basic ones.
Take temporature for example. We can explain temporature in terms of the mean kinetic energy of molecules, but that doesn't mean that temporature is not real. Hardline reductionists might mumble objections that this is only a product of our modes of perception at this point but I think they would be in the minority. Not only that the reduction of temporature is notably non specific! We explain the temporature of gass as mean kinetic energy of molecules and we explain temporature in metals as excitation of free electrons. So temporature is different things in different materials. But it is still real; and it still identifies a set of specific causal properties and propensities. There are many other properties like temporature. Reductionist have difficulty with this.
The mind may be like this. It may be that mental activity is realisable in silicon and tin etc. as much as it is in blood and tissue.
Engineer Joe Michael believes the applications could include clearing landmines, manipulating chemical solutions and optimizing weapon systems -- a clear example of the dangers of emergent behavior.
What is this about? Look out! Studies show that explosives can be fabricated from fertilizer and fuel oil - a clear example of the dangers of gardening.
The naive man accepts life as it is, and regards things as real just as they
present themselves to him in experience. The first step, however, which we take
beyond this standpoint can only be this, that we ask how thinking is related to
percept. It makes no difference whether or not the percept, in the shape given
to me, exists continuously before and after my forming a mental picture; if I
want to assert anything whatever about it, I can only do so with the help of
thinking. If I assert that the world is my mental picture, I have enunciated
the result of an act of thinking, and if my thinking is not applicable to the
world, then this result is false. Between a percept and every kind of assertion
about it there intervenes thinking.
The reason why we generally overlook thinking in our consideration of things
lies in the fact that our attention is concentrated only on the object we are
thinking about, but not at the same time on thinking itself. The naive
consciousness, therefore, treats thinking as something which has nothing to do
with the things, but stands altogether apart from them, and turns its
consideration to the world. The picture which the thinker makes of the
phenomena of the world is regarded not as something belonging to the things,
but as existing only in the human head. The world is complete in itself without
this picture. It is quite finished in all its substances and forces, and of
this ready-made world man makes a picture. Whoever thinks thus need only be
asked one question. What right have you to declare the world to be complete
without thinking? Does not the world produce thinking in the heads of men with
the same necessity as it produces the blossom on a plant? Plant a seed in the
earth. It puts forth root and stem, it unfolds into leaves and blossoms. Set
the plant before yourself. It connects itself in your soul with a definite
concept. Why should this concept belong any less to the whole plant than leaf
and blossom? You say the leaves and blossom exist quite apart from a perceiving
subject, the concept appears only when a human being confronts the plant. Quite
so. But leaves and blossoms also appear on the plant only if there is soil in
which the seed can be planted, and light and air in which the leaves and
blossom can unfold. Just so the concept of the plant arises when a thinking
consciousness approaches the plant.
It is quite arbitrary to regard the sum of what we experience of a thing
through bare perception as the whole thing, while that which reveals itself
through thoughtful contemplation is regarded as a mere accretion which has
nothing to do with the thing itself. If I am given a rosebud today, the picture
that offers itself to my perception is complete only for the moment. If I put
the bud into water, I shall tomorrow get a very different picture of my object.
If I watch the rosebud without interruption, I shall see today's state change
continuously into tomorrow's through an infinite number of intermediate stages.
The picture which presents itself to me at any one moment is only a chance
cross-section of an object which is in a continual process of development. If I
do not put the bud into water, a whole series of states which lay as
possibilities within the bud will not develop. Similarly I may be prevented
tomorrow from observing the blossom further, and thereby have an incomplete
picture of it.
It would be a quite unobjective and fortuitous kind of opinion that declared of
the purely momentary appearance of a thing: this is the thing. Just as little
is it legitimate to regard the sum of perceptual characteristics as the thing.
It might be quite possible for a spirit to receive the concept at the same time
as, and united with, the percept. It would never occur to such a spirit that
the concept did not belong to the thing. It would have to ascribe to the
concept an existence indivisibly bound up with the thing. . .
It is not due to objects that they are given to us at first without their
corresponding concepts, but to our mental organization. Our whole being
functions in such a way that from every real thing the elements come to us from
two sides, from perceiving and from thinking.
The way I am organized for apprehending the things has nothing to do with the
nature of the things themselves. The gap between perceiving and thinking exists
only from the moment that I as spectator confront the things. Which elements
do, and which do not, belong to the things cannot depend at all on the manner
in which I obtain knowledge of these elements.
Man is a limited being... It is owing to our limitation that a thing appears to
us as single and separate, when in truth it is not a separate being at all.
Nowhere, for example, is the single quality 'red' to be found by itself in
isolation. It is surrounded on all sides by other qualities to which it
belongs, and without which it could not subsist. For us, however, it is
necessary to isolate certain sections from the world and to consider them by
themselves. Our eye can grasp only single concepts out of a connected
conceptual system. This separating off is a subjective act, which is due to the
fact that we are not identical with the world process but are a single being
among other beings.
The all-important thing now is to determine how the being that we are ourselves
is related to the other entities. This determination must be distinguished from
merely becoming conscious of ourselves. For this latter self-awareness we
depend on perceiving, just as we do for our awareness of any other thing. The
perception of myself reveals to me a number of qualities which I combine into
my personality as a whole, just as I combine the qualities yellow, metallic,
hard, etc. in the unity 'gold'. The perception of myself does not take me
beyond the sphere of what belongs to me. This perceiving of myself must be
distinguished from determining myself by means of thinking. Just as, by means
of thinking, I fit any single external percept into the whole world context, so
by means of thinking I integrate into the world-process the percepts I have
made of mysel£ My self-perception confines me within definite limits, but my
thinking is not concerned with these limits. ln this sense I am a two-sided
being. I am enclosed within the sphere which I perceive as that of my
personality, but I am also the bearer of an activity which, from a higher
sphere, defines my limited existence.
Our thinking is not individual like our sensing and feeling; it is universal.
It receives an individual stamp in each separate human being only because it
comes to be related to his individual feelings and sensations. By means of
these particular colourings of the universal thinking, individual men
differentiate themselves from one another. There is only one single concept of
'triangle'. It is quite immaterial for the content of this concept whether it
is grasped in A's consciousness or in B's. It will, however, be grasped by each
of the two in his own individual way.
This thought is opposed by a common prejudice which is very hard to overcome.
This prejudice prevents one from seeing that the concept of a triangle that my
head grasps is the same as the concept that my neighbour's head grasps. The
naive man believes himself to be the creator of his concepts. Hence he believes
that each person has his own concepts. It is a fundamental requirement of
philosophic thinking that it should overcome this prejudice. The one uniform
concept 'triangle' does not become a multiplicity because it is thought by many
persons, for the thinking of the many is in itself a unity.
Rudolf Steiner, 1894.
http://home.earthlink.net/~johnrpenner/Articles
Julia, a MUD robot, was often mistaken for a real
i a/
human mudder... It's getting easier to pass the
test when people demand so little intelligence
in interaction on the 'Net...
Find stuff about Julia at
http://foner.www.media.mit.edu/people/foner/Jul
By the way, this is not news, it's just a resurgence
of the HAL syndrome.
Who?