life_intelligence_and_technology
User24 I'm looking for some pointers to places on or off blather where people have discussed the qualities and functions that make up life.

Feel free to start a discussion here.

thankyou.
040131
...
google.com http://www.cogsci.princeton.edu/cgi-bin/webwn?stage=1&word=life 040131
...
u24 Yeah, I was looking more for a.. list of things something has to have or be capable of in order to be classed as alive. EG:

memory
reproduction, with mutation
learning
etc
040131
...
labs.google.com/sets mutation
death
life
learning
memory
Love
perception
logical reasoning
mathematical calculation
Emotion
Other
Language
Personality
Problem solving
040131
...
u24 hmmm... ok

perception and language, mathematical calculation

- OK

but how do you define the others?

logical reasoning
someone on blather defined this as something like
"reduction of all possible outcomes to a series of binary trues or falses"

Emotion
I have no idea how to define this, maybe "a seemingly random influence on logical reasoning?"

Personality
as above?

Problem solving
yeah, but in order to do that, you need to understand the problem. so perhaps communication should come before this?
040131
...
u24 creativity 040131
...
google.com http://baharna.com/philos/life.htm 040131
...
Piso Mojado http://www.amazon.com/exec/obidos/tg/detail/-/0553345842/002-3513229-9632021?v=glance 040131
...
u24 interesting.. "If you had a wiring schematic for a certain person's brain could you read their thoughts?" I don't think so. what matters in understanding how we work is the wiring, which I think is the same for every person (albeit with minor variations (well, OK, sometimes major variations, but these are abnormalities, anyway I digress)) but to know another's thoughts (within the metaphorical context of something with 'wiring') we would need to know the electrical charge of each wire.

:)
040201
...
zeke human life?
what about prions? do you consider them alive?
or are you asking about cognitive life?
040201
...
Norm "Life" is just a word.(http://www.cogsci.princeton.edu/cgi-bin/webwn?stage=1&word=life)

But I don't believe the concept of life can be rightly defined. It encompasses all that surrounds it and changes with those it.

Life requires the consumption of energy.
040202
...
u24 I want to create AI.
so, while it would be easy to create something that simply replicates, I want to create something that learns and makes intelligent choices. I want to create something that is as close to intelligent life as possible.
So yes, cognitive life.
And no, I don't think it can be absolutely defined; such a definition would require absolute knowledge, which we do not possess.
Therefore, I am asking for a definition that loosely encompasses every characteristic that we recognise as intelligent life.
040202
...
zeke The Mind's I : Fantasies and Reflections on Self and Soul
by Douglas Hofstadter

Gödel, Escher, Bach: An Eternal Golden Braid
by Douglas R. Hofstadter

The Engine of Reason, the Seat of the Soul: A Philosophical Journey into the Brain
by Paul M. Churchland

River Out of Eden: A Darwinian View of Life
by Richard Dawkins
040202
...
u24 yeah, I had a look at some reviews of The Minds I (Piso Mojado posted a link above) seems good. But for this purpose, I'm not really looking for discourse, just a simple list.

But yes! I will check out the references given by y'all. eventually.
040202
...
zeke what is your goal? do you intend to achieve the metaphisical and create a being with actual self awareness? or are you attempting to achieve the apearance of intelligence?

Descartes trapped us forever in an infinte regress when attempting to prove that any awareness existed (including his own) using the observable givens. I bring this up because since his work it is very difficult (impossible in the logical and metaphysical sense) to detect intelligence. We can all point to it and basically agree it exists, but proof is based thereafter on a priori assumptions (first principles). in my oppinion, this left the door open to Turing who seemed to be saying that if the construct in question was indistinguishable under laboratory conditions from a person then it is self aware. this seems to be similar to a literal interpretation of the shroedinger's cat conceit or the claim that .9999 repeated is equal to 1. i think that positing a truth that is logical but unmeasurable is arogant.

for me AI is likely to come from experiments with dedicated systems where the operating system is the program and must effect itself as it learns with no backward pathway. massive but finite storage seems to be necessary. i think it would have to mark time in the form of observation of external stimulus, not by it's internal clock (which i think would be like the autonomic systems in us - we are able to sense them but mostly unable to alter them). this begs the question of sensory input. ours is arguably used for survival and ultimatly for species survival. this implys peers, ancestors and progeny.

interaction with other intelligences is how our cognition is activated (mostly). feral children have ostensibly low cognition. maybe AI would require at minimum one peer in order to develop. scarce resources needed for continued existance might be a way to produce a goal oriented cognition.

sorry about the disorder of this blathe, i don't have enough time to do the subject justice. i hope my thoughts are meaningful

see: Turing
040203
...
stork daddy self correcting feedback is a big one. living in an empirical world as we do, it is the ability to navigate such a world in an imprecise and adaptable way, that we use as the benchmark, as our own turing test, for what we consider intelligence. I'm sure you're familiar with work on neural networks in which goal oriented behavior becomes weighted upon reaching the goal. The behavior itself, however, can never be hard-wired, (although in fairness, we have instinctual weightings which are hard-wired - mainly our information gathering) instead the hard-wiring must be in the program by which the behavior is learned. Another sign of true intelligence is the ability to erase behavior, or to allow multiple behaviors based on other mitigating factors. This of course does not have to be an instant process as we all know that habits are hard even for such bastions of intelligence as ourselves to break. Emotion is, of course, more than a random influence. It is a weight, a tendency, a disposition humanity starts with based on thousands of years of trial and error, and varying slightly based on each person's individual lineage and experience. Joy at the sight of loved ones, lust at the sight of the opposite sex, are weights. After all, what is a truly logical decision in a world of infinite decisions? If self-preservation is a prerequisite for something being around long enough to have anything let alone intelligence attributed to it, lust is more than an illogical emotion, rather it is a directive force, a goal, to be placed somehow in the decision tree of outcomes our logical reasoning ability gives us. Where outcomes are true or false. Emotion also allows for incomplete logic. The statement there is nothing in that cave over there, cannot be proven true or false at face value, and yet that situation automatically triggers caution in most humans (unless you are programmed to be the curious or brave type). Another strange property of at least human intelligence which may or may not be neccessary to replicate, is the limited access we have to some parts of our reasoning. The executive branch of consciousness seems to be that which looks over information that has already been processed and then integrates them in a way which leads to a decision. The details behind this are still not understood, as decisions can happen both quickly and slowly and still seem as opaque to both outside observers and the person themself. Back on the topic of intuitive logic though, behaviors could start out with certain weights in an approximation of evolution, but for most accuracy in mimicing true intelligence should be put in an envoirment which demands adaptive behavior as a component of fitness. THe problem with this is we may not understand the changes that are occuring even when we understand the initial learning mechanisms. True intelligence has black box function. It works outside, and is hard to see inside. After all, if people understood the human mind more rigourously i wonder if they would discuss AI like it's such an impossibility. just some random thoughts. all of the above books are good reads by the by. 040204
...
realistic optimist nanotechnology
snow_crash
life_expectancy
analogue computers
040204
...
zeke very well said. your fluency and ideas on the subject are impressive. especially your ideas about sensory feedback loops and the brief reference to intuition. i think extrapolation is perhaps a goal and gestalt pattern recognition. we seem to have hard wired gestalt switches in the systems which lead to fight or flight. also the cognitive enhancement of imperfect sensory input such as edge detection for sight and hemispheric comparison for directional depth perception in audio signals. aproximate cross references such as the strong association between smells and memory (and sometimes the spurious smells generated by the cognitive process in disassociative syndromes).

you have inspired me to think.
040204
...
stork daddy there's certainly a lot to be learned in how our minds make the most of fuzzy edges. i find the most interesting experiments are the ones which show where our perception and intelligence often fails. 040204
...
stork daddy and certainly it is interesting how certain memories code for one another, so that, like you said, they can be cross-referenced as the most relevant association to make with any given sensation. 040204
...
zeke signal and noise. i think that the noise inherent in any senory input system is worth examining. i have begun to wonder if dreams consist of short term memory storage of noise produced by sensory systems during unconciousness which is interpreted using comparisons to patterns stored in long term memory of stimulus actually experianced. this interpretation carries the associative meaning and symbolic value over to the new "memory". this might explain why in most dreams the "narritive" and subjects of the dreams seem so random. they are. in this model a dream is actually experianced as you remember it not as it happens. it is post processed. 040204
...
stork daddy interesting. now would this serve a function or would it rather be an accidental or tag-along property of the brain's methods for assigning meaning to sensory noise. it would certainly explain why dreams seem in some way less real. although lucid dreams would need explanation. 040204
...
zeke i am not sure, it is just a supposition. not even a proper theorem. i am inclined to think that it is part of the system that cuts out the physical part of our reflexive behavior when we sleep. giving a memory like structure to formless sensory material might allow sleep to happen uninterrupted. the material accumulates and needs to be understood at the experiential level because that is what those systems do. 040205
...
u24 hmm.. agreed, stork, zeke's is an interesting theory, especially regarding the post processing idea.

OK.

Simply beating a turing test is a relatively easy target. Especially if I cheat. But while it would be interesting to create something that could mimic us, it wouldn't serve any purpose (for me - obviously the commercial applications would be endless, but that's not my goal)

no, initially, I want to create something that learns from it's mistakes, behaves fairly unpredictably, and swaps it's memory with other programs. From that point onwards, I'd have to see where it's greatest failings were, and improve upon them. Ideally, the goal is to create a self aware program, but whether or not that is possible (and whether I'd be able to tell if it was self aware or not) I don't know.

I agree, 'the Turing test', while interesting, is by no means a measure of intelligence or self awareness (NTS: is there a distinction?)

Why do you think massive storage would be neccessary? Of course, if it was to store the same amount of information as us, and interact with our environment, then yes, I can see why. But what I had in mind would, at least to start with, interact with a virtual environment - otherwise I'd have to waste time coding pattern recognotion, as you say, edge detection, colour matching and a whole host of other things, before the program would even be able to see/hear/etc around it. In a virtuial environment, I can simply provide direct stimuli - go here and die, go here and start getting less healthy, go here and start getting more health.

In a simple environment such as the above described, the program would be hard wired to like being alive and healthy. Then, without actually having pre-programmed moves, it would move to health+ squares if it happened across a health- square. It would remember that square#4 was a health- square, and not re-visit it.

Of course, this is very simple decision making, but each time it did what I saw as the correct choice, I would add another set of environment rules, and test out behaiviour patterns until it seemed ok with that environment.

I'll to knock out a rough beta within a month.

you could argue that such a program could never be classed as intelligent, as it would only be able to interact with the environment in ways I allow, but in the same way, we are only able to interact with the environment in ways that science will allow.
Also, such a program would not be self aware, merely a choice maker. But I think it would be a good start.

Re marking time; yes, we are ruled by external factors (known as zeitgebers in psychology) If a computer were to interact with our environment, I agree, it would need the same input.

And yes, interaction with others of it's speices would be essential. Whether it would breed or merely continously evolve is uncertain- breeding implies death, whereas evolution implies a single entity - both are not ideal circumstances. I would like to see a lot of programs running around and merging with each other to create new, blank programs, but on the other hand, it would be easier to create something which simply continually adapted.

And it would be very interesting to see how the programs interacted.

The 'scarce resources' idea is what I was trying to say above with the health+ and health- areas of a virtual environment. But I think you said it better. So I'm stealing that phrase. :)

Ok, stork daddy;

"navigating the world in an imprecise way"

Exactly! One thing I've been keen to avoid is creating a program that plays perfect chess. In other words, that goes through every possible combination of decisions from this point onwards - Such a species would be perfect, always making the correct/best choice, but would be totally predictable, While it would be interesting to see how it's choices changed as it learnt new things about it's environment, I don't think such an entity would be very interesting to study. I want to be able to look back and say "I have no idea why it did -that-!" - as humans, we are obviously imperfect, and that's what I want to reproduce, not a perfect entity that never fails. Where's the fun in that?

And yes, adaptation is important - the program breifly discussed above would learn from past experience, in effect, at each stage, it would say "Have I been in this situation before?", look at what it did, what happened, and weigh up whether this was a good decision, or whetehr the chances are that making a random decision would work out better.

the Hard Wired behaivour would be as minimal as possible. it would be able to 'see' it's environment, 'feel' pain when it loses health, etc, however I was thinking of randomly deviating the hard wired weightings each time a new program was born. EG:

the perfect program would have +10 feel good points when it recieved food. Therefore it would enjoy eating food.

But each time a new program was created it would randomly change it's weighting by up to 5 points, hence, some programs would have +15, and -really- like eating food, and some only +5. This way, I think we can mimic (possibly grossly more/less than we need to) the way each human has unique weightings.

And yes, I beleive the program would have to develop habits, it may learn that in addition to being hard wired to like food, that it usually ate food in order to live, and therefore it would raise it's weighting on food. I'm not sure exactly how this would work, I think the best way would be to create the simpler versions and then expand on them.

Yes, human emotion is moer than random, and can sometimes even be explained. But it's effect is to randomise the logic, hence, the program would firstly work out the most logical decision, taking into account it's own weightings, then randomly mutate that decision, so that there's a small chance that it would not make the most logical decision. This process can be fine tuned along the course.

It would be interesting to create a program that realised it's own weightings were not the most beneficial, and was able to change them. a metaprogramming program. (!)

The program would live in a turn based environment, where each turn represents one decision, and it would give a report on why it made the choice it did.

I would also be able to look at it's memory to see what it has learned.

Phew. thanks for responding. :)

And btw, what's a Gestalt switch?
040206
...
stork daddy don't forget vicarious learning 040206
...
z gestalt switches: the seeming abillitiy of almost autonomic decision making in a fight or flight situation for enhanced survival (if you allow a bit of poetic licence). 040206
...
zeke user24, aside from your references to adaptation, learning and evolutionso far what you are describing sounds like a complicated Tomagachi. 040207
...
nom (read these yet?)
if not check out:

regarding prions and memory-

http://www.eurekalert.org/pub_releases/2003-12/wifb-cm121703.php

recent Wired article:
Living Machines-

http://www.wired.com/wired/archive/12.02/machines.html
040207
...
notme maybe you already know about?:

kurzweilai.net

how to build a brain-

http://www.kurzweilai.net/meme/frame.html?m=3


in the news:

Nanotube quantum dot doubles up-


http://www.nanotechweb.org/articles/news/3/2/2/1
040207
...
u24 cheers for the links, nom I will check out. (what are you, anyway, some kinda link-o-matic? :D )

and zeke, re tamagochi... well... yeah!
but this is just the beginning - as I said, it's just a beta (or will be) to see how easy it is to get it to make decisions without pre-programmed choices. From that ponit on, I'll have to see what the next logical step is. But yes, initially it won't be very impressive.. more like a furby, though, I think, once I've built the ability to communicate with other progs.

btw, someone remind me to actually do this at some point.
040210
...
Lick Are emotions necessary for artificial intelligence? 040210
...
zeke what are emotions? 040210
...
zeke http://www.wired.com/news/gizmos/0,1452,62184,00.html 040211
...
zeke heuristic AI 040221
...
z user24: btw, do this at some point. 040221
...
u24 thankyou. I did start (got about half way), but I lost heart a little (as usual).

I'll finish at some point.

thankyou for reminding me :)
040310
...
z you are welcome. 040311
...
u24 I think it was a bit ambitious of me; I actually know nothing about existing AI systems. (do I re-invent the wheel? yes. I also tend to re-invent the road.) 040402
...
u24 ((then I get bored and start wondering about hovercraft)) 040402
...
notme i want a hoverbarrow 040402
...
parrot sketch birdmad my_hovercraft_is_full_of_eels 040402
...
marked . 050120
...
z http://y.20q.net/ 050202
...
Nathan Solla According to my Biology instructor, Life, in order to be classified as "life" must do ALL of the following:

1. Begin. (birth, generation, germination, mitosis, etc.)

2. Grow and Develop. (Change over time, age.)

3. Reproduce itself. (Babies, offspring, eggs, spores, splitting in half, etc.)

4. Consume Fuel for Energy. (Eat food)

5. Need Water to survive.

6. Excrete Waste. (Poop)

7. Contain Cells. ("The_building_blocks_of_life...")

8. Die. (End, Decompose)
050203
...
u24 need water? that's a bit specific.. 050203
...
z is a prion alive? 050203
...
u24 surely #3 should involve random mutation?
do electrons count as fuel?
does removing unused variables count as excretion?
do functions count as cells?
are mules alive? (they cannot reproduce)
050719
...
Lemon_Soda Gosh this is silly.

Everything is alive. Even inanimate matter. Everything moves. Everything requires this set of "food" to be produced. Everything changes(mutation, and when its done being what it is it leaves behind what it doesn't need to be this new thing, or waste.) All things need "shelter" to keep from "being killed" which is just ceasing to be in its current state. Everything reproduces, just not in the way we normally associate with the term. If it didn't, were did everything come from? Even stars "die". As for intelligence, I think everything is equipped with exactly as much intelligence as it needs for what it is. What good is a human intellect to a rock? Could an elephant run on a jellyfishs nervous system? And finally technology. Nature has been doing THAT alot longer than we ever started.
050719
...
Lemon_Soda Truth be told, there is nothing BUT life. Death is just a grim term for entropy, the natural process of lifes movement.

is having a Yoda moment.
050719
...
u24 but you understand nonetheless that the idea that 'everything is alive' is controversial.

why do you think this is so? why is there traditionally perceived a distinction between life and non-life*?

and, given that there is a traditional distinction between the two states, what defines the boundaries of the aspect of the fluxlike whole that we traditionally term 'life'?

*rather than death, which implies ex-life

:-)
050719
...
Lemon_Soda The traditional belief in life and nonlife is tied to a beings inability to comprehend beyond what it is. Someone grew old and died and someone else witnessed it. THe dead was nolonger human, but a corpse, food and microorganisms, an entirely different form of life but still life. The Witness did not think about that. The witness thought about how the person who was just walking around isn't any more and never will again. Because of the transition, the witness understood that the type of life this person represented ended in this case, or died. But because we didn't have microscopes and science and so on for so many thousands of years dead meant dead, not new life in a different form. However, idea wasn't completely alien to us that everything had its place and that life continued on all scales. The great circle of life and all that is just analogy for the whole process. The modern and pre modern view of death as non-life is a bit askew because coninuing as life other than in the concious, human state that we enjoy, is too alien and undefinable for us to except without fearing it. I have no idea what its like to be a mass of meat slowly being digested and turned into different components of the next form of life. I don't want to know. I guess I'm greedy like that. 050719
...
anne-girl dude, this is fascinating

can't just make something which emulates intelligence (millions of if-then statements) - like randomness can look like intelligence, but it's not
and statistical analysis (bayesian...) can do similar things, but it's still not intelligent

what defines intelligence? sentience, self-awarenesss
doing the unexpected, maybe?
like a program that did something it wasn't supposed to do... like randomly commit suicide (or the AI equivalent of suicide)

i guess it'd have to have free will or something

how to implement that? gah.
i mean... being random and irrational i think is characteristic of being human, where bits and bytes and rationality is all computers can do
(it seems that the best || most usable technologies (for sorting information, etc) are the ones that put human intelligence && computer number-crunching powers in good ways)

i don't actually know what i'm talking about... just some thoughts
050719
...
marked . 051116
...
u24 wrote a 4200 word essay on "Can machines be conscious?"

it's not that great though.
I'll post it online at some point.
070110
...
z please do. 070110
...
They call me Truth Good God i have a lot of reading to do! 070630
...
z spore 080923
what's it to you?
who go
blather
from