Artificial Intelligence: Why Would We Make One?

[Lord of the Abyss]

Okay you killed the useless part of the population... however the fewer people requires less products (even if the unemployed consumed a small amount, they consumed some) which means there is a decreased need of manhours. After firing the now useless employees their consumption dramatically decrease, which initiates a new round of workforce reduction...etc. So within a very short amount of time the original situation (40% working 60% unemployed) is recreated only with a smaller population base (and a bit of oversized robotic/AI workforce since it was designed for the original size of the society).

And? In that scenario there isn't anything to keep the elite who control the robotic army from systematically killing the "useless" members of the population until there's nothing left but themselves. They've got the robots to support and defend them, so the extermination of the other 99% of the population won't cause them any inconvenience.

[someone_else]

So first you observe it, then you theorize about it, THEN you test that theory in this case by making a computer that thinks like it

You keep trumpeting with your ass.
Human brain (or any other kind of modern science for that matter) is too complex to do this process on the whole thing at once. You have to split the subject and make lots of different theories about how all the tiny pieces (small enough that you can work with them) work and then, when all the smaller theories are tested and set down, then you can make unifying theories on how the wole thing you are studying works.
(in reality there are more more sub-divisions of subjects and sub-unifying theories though)

You're basically saying that people can try to formulate theories about how human body works as a whole without knowing a shit of uhhh... cellular biochemisty or fluid physics. Sure, you can do it, but it is kind of grossly inefficient.

So, what I was trying to say, is that to test each sub-theory you make an algorythm, when testing unifying theories you have to link these algorythms and look at the results. If you link enough algorythms to test unifying theories you get something like the AI described by the OP.

If it thinks like a human, emotes like a human, has a personality like a human, all right down to the functioning of its brain...

You're jumping to the end of the research process. For more or less all the research time, the thing is a computer programmed to emulate how a part of the human brain works at the best of the creator's understanding, built to increase this understanding.
One of the main benefits is working only in a virtual world where you can accelerate the "time" of the simulation, so that a research that in real life done with humans would take decades in this way takes very limited periods of time.
This virtual environment doesn't become something like Matrix until you start to work on something close to AIs as described by OP.

Anyway, as long as the machine doesn't react to each and all situations within the statistical variability of the reactions of a human it doesn't qualify as "human" (although can qualify as "alien" with ease), thus you can use it for research.

This means there will be huge "gray areas" and you don't know exactly when you start to cross the line and create a sapient being until after you did it.

Then, when the first AI as described by the OP is born and noone kills it out of fear, you can have such kind of synthetic creatures whose life serves no pratical purpose due to huge reliability concerns.

So, what I was trying to say, is that imho the only way to have AIs as described by OP is as a byproduct of research in how human brain works.

Noone wants a sapient weapon that outsmarts the gunner (and the entire command staff as a whole).

Furthermore, even if it wasn't, you are proposing experimentation on a being intelligent enough to understand its own existence, and form opinions on what should be done with itself

This would happen only at a very advanced stage. As I said, you neet to test loads of sub-theories about very specific aspects before, and for those you aren't dealing with a full-grown AI, but only with very specific parts of brain.
Then you can have Evil Scientists that go further (you always have some), FOR SCIENCE!!! and get an AI like the OP wants.

That or the theory has no basis in reality because the actual machinery under the hood is completely different from a human brain.

Why? It is designed to work like a brain and not like a computer (it is there for research, not for processing power). Each nervous cell works by giving a signal if receives enough signals to be activated and not enough signals to shut up, and I don't see that particularly complex to code.
If you manage to roughly recreate the right connections between a ludicrous amount of such "virtual neural cells" (quite a feat), you should in theory make "virtual parts of brain".

The neural pathways that make up the computing basis of the brain are constantly rewiring themselves

It is part of the learning process, and as such it must be done by the computerized model too. As long as such rewirings happen for a reason and they follow some kind of logic, it can be emulated.

Even once you have a model, its not going to work for everyone, and it will eventually be rendered moot

Don't make me laugh. Not even sleeping pills work well for everyone. The point is making models that work within the results of statistics on humans.

Which makes everything you discover entirely artificial and devoid of real world application.

Starting from a blank slate allows you to see where the differences are between your ideal model and the real life human. Thus in what direction new theories must go to explain why it didn't work as a real brain. Theories get tested by experimentation and if they work, then you improve the model and discover what has an effect on human brain as well.

[Starglider]

And? In that scenario there isn't anything to keep the elite who control the robotic army from systematically killing the "useless" members of the population until there's nothing left but themselves. They've got the robots to support and defend them, so the extermination of the other 99% of the population won't cause them any inconvenience.

In practice I don't think people are that amoral. Even if you think a substantial number of ultra-rich are, which I'd dispute, actions like that still require the support of robot designers and most likely the military. The 90% would be rounded up and contained in welfare slums, although they might be quite pleasant welfare slums by contemporary standards, depending on how much production was allocated to them.

Not a completely hypothetical scenerio as you could do this without needing general AI (and thus without recursive self-enhancement rendering the desires of the owners irrelevant), although when you have robots this flexible general AI won't be far behind.

I missed the nonsense from Formless earlier, but obviously many thousands of academics vastly more qualified than him have devoted their careers to building computational models of the brain purely for brain research purposes, which does tend to suggest that he is spewing bullshit like a firehose accidentally connected to a sewage farm.

[someone_else]

Okay you killed the useless part of the population... however the fewer people requires less products (even if the unemployed consumed a small amount, they consumed some) which means there is a decreased need of manhours. After firing the now useless employees their consumption dramatically decrease, which initiates a new round of workforce reduction...etc. So within a very short amount of time the original situation (40% working 60% unemployed) is recreated only with a smaller population base (and a bit of oversized robotic/AI workforce since it was designed for the original size of the society).

And? In that scenario there isn't anything to keep the elite who control the robotic army from systematically killing the "useless" members of the population until there's nothing left but themselves. They've got the robots to support and defend them, so the extermination of the other 99% of the population won't cause them any inconvenience.

Why should the AIs obey if they are made similar to a human mind (as the OP suggested)? Also, what prevents the AIs from seeing us as different (which is a given for any human mind looking at another) and killing us all thinking they are The New And Better Men?

[Starglider]

Why should the AIs obey if they are made similar to a human mind (as the OP suggested)?

Patching a simulation or near-simulation of the human brain to seem obedient isn't too hard. You've got direct access to all the reward centers, you can directly map reinforcement / which network sections are modified by learning, you can isolate sections of the NN and run tests with millions of stimuli to check that the outputs are appropriate. You can tack on as much external watchdog code as you want, performing black box compliance tests (e.g. recognising and executing an order to shut down) and white box AI state monitoring (for dangerous cognitive patterns). Plus even a selfish AI is going to pretend to be obedient under test conditions if it's smart enough to realise that this is how to be allowed out into the real world.

Of course 'seem' is the key word there. No such messy, 'adversarial' goal-system design could possibly be stable in the long term.

[cosmicalstorm]

A quick question for you Starglider, while you are around. What do you think about sci-fi people who dream of galaxies filled with humans who travel back and forth between the stars in giant human friendly spaceships? If we do not self destruct, is that very realistic?

[Starglider]

A quick question for you Starglider, while you are around. What do you think about sci-fi people who dream of galaxies filled with humans who travel back and forth between the stars in giant human friendly spaceships? If we do not self destruct, is that very realistic?

Obviously I'm not qualified to answer specific questions about interstellar travel etc. I have deep specialist knowledge of a couple of things, artificial intelligence and software engineering, so I try to answer questions about those fields and point out any obvious uninformed crap I see (which is a lot, for AI). If you're interested in the specifics of physically plausible interstellar travel / spacecraft / long-term confined environment psychology then ask the astrophysicists, engineers or social sciences people we have in abundence on SDN.

Since I think we're likely to create AGI in the near future and that it will quickly be ridiculously powerful, my answer is unsurprisingly 'if we solve FAI then we will most likely be able to do that if people still want to, if we don't then everyone will be dead before we get the chance'. I have no idea how popular travelling around in spaceships will be to a post-singularity population, I'm pretty sure some people would want to do it and some fraction of them will be prepared to do it the 'old-fashioned / retro way', despite the inconvenience. Whether that fraction will be significant I don't know. It's interesting to compare the two Greg Egan novels 'Diaspora' and 'Schild's Ladder'. Both of them have hard science interstellar travel, both of them have sentient AI (only slightly transhuman due to understandable author limitations) and mature nanotechnology (femtotechnology being independently developed in both novels ). The difference is that in Diaspora the number of fleshy humans is so small, and the people who want to be fleshy humans tend to be interested in living on earth in vaguely naturalist eco-communities, that no one bothers investing the resources in slow fleshy-human-supporting interstellar ships. Amusingly there's a part of the novel featuring uploads travelling on an AI ship (essentially a 1m spherical mainframe on an antimatter rocket) who are experiencing the journey as a simulation of a much bigger ship crewed by fleshy-genetically-engineered-human-derrivatives (as an entertainment). In Schild's Ladder the number of fleshy humans is significant; both the ones with implanted electronic brains who do regularly data-teleport themselves between stars, but who also like to have physical starships, and the luddite 'real human' group who insist on travelling by giant fusion torch-ship (because they think uploading eats your soul). Both scenarios are possible, but as I said, only if we build FAIs that allow it.

[Broomstick]

AI would be a technology like cell phones or the internet. Care to demostrate how either of those has widened the poverty gap?

Outsourcing.

Case in point: In my old job used to spend 1/4 of my time retrieving research from medical libraries. Shortly before I was laid off that was outsourced to India and is now performed entirely over the internet. I used to spend about another 1/4 of my time making travel arrangements for people. That is now done over the internet. That's half my former work right there. This sort of outsourcing was a major reason my job was eliminated and there are now no longer many such jobs around. As the loss of that job was the reason I went from Western Middle Class to Poor I'd say that in my case, yes, the increasing use of the internet "widened the poverty gap".

Repeat for several millions others in the western world.

Now, yes, there will probably be other jobs that develop, but there's no guarantee I will have the education/skills for those jobs, nor the means to acquire them. Meanwhile, 5-10 years of poverty while all this sorts out totally sucks, and will significantly impact my ability to provide for my old age.

I'm sure around 1910 the buggy whip makers faced similar issues.

The fact is, when machines take over a type of work from human beings there is at best a very uncomfortable period of adjustment. Sometimes, it can result in long-term misery in human beings. Recognizing that effect does not make one a Luddite (indeed, from what I've learned about steel mills from living in this area I don't think anyone is arguing against increased automation, even those hurt by it because a steel mill is a little slice of hell on earth) but one really does need to recognize that the benefits of automation really do come with a cost.

[Broomstick]

In practice I don't think people are that amoral. Even if you think a substantial number of ultra-rich are, which I'd dispute, actions like that still require the support of robot designers and most likely the military. The 90% would be rounded up and contained in welfare slums, although they might be quite pleasant welfare slums by contemporary standards, depending on how much production was allocated to them.

Not in this country, where the poor are to be punished for being poor. While some nations might keep the unemployed masses in a gilded cage others will consign them to essentially a prison existence.

[cosmicalstorm]

A quick question for you Starglider, while you are around. What do you think about sci-fi people who dream of galaxies filled with humans who travel back and forth between the stars in giant human friendly spaceships? If we do not self destruct, is that very realistic?

Obviously I'm not qualified to answer specific questions about interstellar travel etc. I have deep specialist knowledge of a couple of things, artificial intelligence and software engineering, so I try to answer questions about those fields and point out any obvious uninformed crap I see (which is a lot, for AI). If you're interested in the specifics of physically plausible interstellar travel / spacecraft / long-term confined environment psychology then ask the astrophysicists, engineers or social sciences people we have in abundence on SDN.

Since I think we're likely to create AGI in the near future and that it will quickly be ridiculously powerful, my answer is unsurprisingly 'if we solve FAI then we will most likely be able to do that if people still want to, if we don't then everyone will be dead before we get the chance'. I have no idea how popular travelling around in spaceships will be to a post-singularity population, I'm pretty sure some people would want to do it and some fraction of them will be prepared to do it the 'old-fashioned / retro way', despite the inconvenience. Whether that fraction will be significant I don't know. It's interesting to compare the two Greg Egan novels 'Diaspora' and 'Schild's Ladder'. Both of them have hard science interstellar travel, both of them have sentient AI (only slightly transhuman due to understandable author limitations) and mature nanotechnology (femtotechnology being independently developed in both novels ). The difference is that in Diaspora the number of fleshy humans is so small, and the people who want to be fleshy humans tend to be interested in living on earth in vaguely naturalist eco-communities, that no one bothers investing the resources in slow fleshy-human-supporting interstellar ships. Amusingly there's a part of the novel featuring uploads travelling on an AI ship (essentially a 1m spherical mainframe on an antimatter rocket) who are experiencing the journey as a simulation of a much bigger ship crewed by fleshy-genetically-engineered-human-derrivatives (as an entertainment). In Schild's Ladder the number of fleshy humans is significant; both the ones with implanted electronic brains who do regularly data-teleport themselves between stars, but who also like to have physical starships, and the luddite 'real human' group who insist on travelling by giant fusion torch-ship (because they think uploading eats your soul). Both scenarios are possible, but as I said, only if we build FAIs that allow it.

Thanks for the reply. I often enjoy the posts you make here. I wasn't curious about the possibility of interstellar travel in itself (which has been discussed in length on here and elsewhere) only the notion of how likely it is that humans very similar to those alive today will be sitting inside ships heading to the stars. (A situation which a lot of sci-fi authors still seems to consider very likely.)

[Formless]

So first you observe it, then you theorize about it, THEN you test that theory in this case by making a computer that thinks like it

You keep trumpeting with your ass.
Human brain (or any other kind of modern science for that matter) is too complex to do this process on the whole thing at once. You have to split the subject and make lots of different theories about how all the tiny pieces (small enough that you can work with them) work and then, when all the smaller theories are tested and set down, then you can make unifying theories on how the wole thing you are studying works.

So fucking what? The same is true of ecology, dumbshit. And yet simulations of ecosystems require data observed in nature to make predictions-- not the other way around, and they can easily miss variables that need correction. Ditto for the weather. Or for the economy. Or any other complex system you want to study. You can't just rely on modeling.

You're jumping to the end of the research process. For more or less all the research time, the thing is a computer programmed to emulate how a part of the human brain works at the best of the creator's understanding, built to increase this understanding.

Hey, dumbass, guess what: real human brains are too interconnected for this approach to work. Do you know anything about neuroscience? Anything at all?

One of the main benefits is working only in a virtual world where you can accelerate the "time" of the simulation, so that a research that in real life done with humans would take decades in this way takes very limited periods of time.

Riiiiight. Somehow we're magically going to be able to compute all the possible interactions of a network that has trillions of connections and trillions more possible connections faster than we can simply observe it in action in real life. This kind of wishful thinking just never seems to get old with techno-nerds, does it?

This virtual environment doesn't become something like Matrix until you start to work on something
close to AIs as described by OP.

Anyway, as long as the machine doesn't react to each and all situations within the statistical variability of the reactions of a human it doesn't qualify as "human" (although can qualify as "alien" with ease), thus you can use it for research.

People have a problem doing experiments on rats for fucks sake! Again, have you ever read an ethics textbook in your goddamn life, you smug little cretin?

Why? It is designed to work like a brain and not like a computer (it is there for research, not for processing power). Each nervous cell works by giving a signal if receives enough signals to be activated and not enough signals to shut up, and I don't see that particularly complex to code.

We've gone over this kind of thing in the past. Wyrm described it best in this thread:

A brain is not a neural network as defined in computer science textbooks. It is a physical machine of action potentials floating in a modulating broth of fluid. That's not amenable to translation into logic functions.

that's just one example of how a model and the actual thing being modeled can significantly differ.

BTW, I highly suggest you read that thread, especially Alyrium's posts earlier on in it.

The neural pathways that make up the computing basis of the brain are constantly rewiring themselves

It is part of the learning process, and as such it must be done by the computerized model too. As long as such rewirings happen for a reason and they follow some kind of logic, it can be emulated.

The problem is that for it to give real world predictive results about how humans behave it must simulate not just the brain but the environment that the brain lives and develops in. That adds to the total number of things you have to simulate, and brings in the possibility that there are unobserved variables you didn't simulate, at which point you might as well just observe naturally born humans.

Even once you have a model, its not going to work for everyone, and it will eventually be rendered moot

Don't make me laugh. Not even sleeping pills work well for everyone. The point is making models that work within the results of statistics on humans.

Which is why I say you might as well throw this idea in the trash because to get statistics on humans you have to do population studies. You really have no idea how the social sciences work, do you.

Starting from a blank slate allows you to see where the differences are between your ideal model and the real life human. Thus in what direction new theories must go to explain why it didn't work as a real brain. Theories get tested by experimentation and if they work, then you improve the model and discover what has an effect on human brain as well.

Do you have a reading comprehension problem? I already said that there is no "ideal human" you can model, shitwit. That's why real social scientists do population studies whereas techno-nerds think you can model the brain like a solar system. You can't.

I missed the nonsense from Formless earlier, but obviously many thousands of academics vastly more qualified than him have devoted their careers to building computational models of the brain purely for brain research purposes, which does tend to suggest that he is spewing bullshit like a firehose accidentally connected to a sewage farm.

There are respected physicists tho think they are experts on biology too, but that doesn't mean they are. Its possible to be an expert at one thing and also be an arrogant moron who knows not the limits of ones own expertise. To my knowledge, all the computer scientists who are studying things like neural networks and the like do it precisely to better understand their own fields, not neuroscience, neurology, or psychology.

[Samuel]

It'd be more than manual labor (which itself isn't just digging ditches and the likes). Think of every form of skilled profession that requires hands-on involvement, from doctors to plumbers. If anything, we'd need a lot more such people, to maintain and do upkeep on all the extra infrastructure for the many AIs we'd have running. Not to mention all the medical and personal assistance our gigantic population of retirees/old people are going to need.

It kind of reminds me of this hypothetical essay by Paul Krugman, written back in the mid-1990s (although he doesn't factor in potential AI involvement in design/operations).

Doctors can be replaced by robots. We already have machines to do surgery and rules to be used to diagnosis . I guess we might have more "skilled" versions of labor and service personal, but that isn't going to make up a large part of the economy. We aren't going to need alot of workers to maintain the "extra infrastructure"- just run the AIs on regular computers and toss the modem when it stops working.

This doesn't require amazing robots. Asimo type robots would be fine, slightly scaled up. Sure they cost ~$1,000,000 USD to make at the moment, but a Chevy Malibu probably would too if we built less than 100 cars a year, globally. Mass-produced cost-effective (in the sense of being cheaper to lease than a human worker's salary) robots capable of at least light human labor are already possible, we just don't have the software to make them useful.

Or you take advantage of the massive glut in manual laborers due to everyone lossing "knowledge-worker" jobs and drive wages down. You can play feudalism if you are meglomaniac enough- get bunch of workers, find all the outside inputs you need and replace them with labor as much as possible and pay your workers in food, clothing, housing, medical care, etc. You can get them to work less than the minimum wage if you drive down the cost of goods enough and you can have them as "volunteers" or some other eumphenism to avoid legal sanction.

I think this is alot more likely than welfare slums. Sure, you can be super rich and hang out with your rich buddies in magnificent wealth. But there are more things in life than that. Building monuments, living in a castle, having followers to make you feel important, surrounding yourself with beautiful men and/or women who will do anything you want, having enough supporters so that you don't even need to pretend to be in a democracy... all these things could be yours.

[Broomstick]

Thanks for the reply. I often enjoy the posts you make here. I wasn't curious about the possibility of interstellar travel in itself (which has been discussed in length on here and elsewhere) only the notion of how likely it is that humans very similar to those alive today will be sitting inside ships heading to the stars. (A situation which a lot of sci-fi authors still seems to consider very likely.)

Any space craft which would allow the current variety of H. sapiens to travel interstellar distances without FTL travel would have to be a self-contained, independent eco-system. A self-contained world. If you have such a spacecraft a legitimate question is "why would you want to land on a planet?"

Possible? Perhaps. But it's a different question than "Why would someone want to build an AI?" I don't see where an AI would be a necessity for such a spacecraft, though it would probably make running one easier.

[cosmicalstorm]

As for the brain emulation argument I can really recommend the Whole Brain Emulation Roadmap by Sandberg and Bostrom. It seems to cover most of the points taken up in this thread so far. Suffice to say there does not seem to be any particular reason for why it shouldn't be possible to do it. But it will be really hard. (Perhaps a true AI will have been developed before this project is done)
http://www.philosophy.ox.ac.uk/__data/a ... report.pdf

Possible? Perhaps. But it's a different question than "Why would someone want to build an AI?" I don't see where an AI would be a necessity for such a spacecraft, though it would probably make running one easier.

I was aiming more for the idea, that by the time we are capable of designing such ships (decades, centuries and possibly never due to society prioritizing other things), we will have long since entered the transhuman age, or succumbed to some global disaster. I.e there will never be a ship filled with standard humans as we know them, where computers only act as helpful tools for navigation.

[someone_else]

You can't just rely on modeling.

Of all your post, in this only sentence you are damn right.
But I don't think I did say we should use modellling for everything, only that advanced models of human brains (that have some kind of use already) could become somewhat similar to AIs as described in the OP as they become more and more complex.

Hey, dumbass, guess what: real human brains are too interconnected for this approach to work.

That's why I said starting small and going big with time. It will take centuries of reserach and also advances in computers too. It has also the side-effect that when you start to work on a model that blurs the line with AI you have a semi-perfect understanding of human brain.

Somehow we're magically going to be able to compute all the possible interactions of a network that has trillions of connections and trillions more possible connections faster than we can simply observe it in action in real life

Not exactly, I said we could make a model that mimicks the workings of a human brain and shape its "growth" so that at the end the reactions are within human variability. It's not anywhere near the same as predicting the outcome of all the interactions, although the results are similar.
It grows as we gather more info about how the human brain works, by making theories about why the model doesn't work as a human. It isn't a complex concept to understand, I think.

People have a problem doing experiments on rats for fucks sake!

That's because rats are cute, warm and cuddly. Noone gives a fuck for experimentation on insects, or on cell cultures for that matter. (unless they are GASP-OMG human embryo staminal cells, of course since "God doesn't want it")

have you ever read an ethics textbook

No. Do ethics textbooks exist at all? All I read on ethics was on philosophical textbooks.

A brain is not a neural network as defined in computer science textbooks. It is a physical machine of action potentials floating in a modulating broth of fluid. That's not amenable to translation into logic functions.

While written in a professional and convincing style with the correct buzzwords (to the contrary of your street-dirt ramblings), the bolded sentence is pretty idiotic. Pretty goddamn fucking unbelieavably idiotic.
Regardless of how the brain works (and of our limited understanding of it), it follows specific and unchanging (if pretty complex) rules just like any other natural phenomenon.
Computers are machines made to follow rules, to excel in following rigid rules.
Computers can in fact be programmed to do incredibly stupid and wasteful number-crunching that have little to do with their architecture like making realistically-rendered 3D environments on a flat 2D screen for totally pointless shooter games, or predicting weather patterns, or predicting if a vehicle frame or shape would work in theory or not.
If it is programmed to work like the current understanding of a human brain, it will follow the program, but will probably waste a lot of its processing power to emulate the brain since the architectures differ quite a bit.
I'm not saying that it is easy (it is indeed mindboggingly complex, but possible).

That adds to the total number of things you have to simulate, and brings in the possibility that there are unobserved variables you didn't simulate

Yes! That's exactly the point!
Seeing where the model doesn't react like a human brain will allow you to detect that there are variables that may have been left out. And so you can start looking for them.
How can you make make experiments to look for variables you don't even knew existed?
Divination? Illumination? SMS from Flying Spaghetti Monster?
That still implies that you have statistical data about how humans with a similar background usually react.

Do you have a reading comprehension problem? I already said that there is no "ideal human" you can model, shitwit.

You should not talk of things you clearly don't understand.
The "average" of population studies is usually the "ideal human" most other medical science experiments assume they are dealing with, and the "ideal target" of the testing of most medicines. It's the reason why not all medicines and procedures work well on all people and there are different medicines (not just different brands, that's another story) to cure the same illness.
There are always differences between theory and practice, but the more advanced your knowledge gets, the less these "unexplainable rare differences" there are.

[Broomstick]

People have a problem doing experiments on rats for fucks sake!

That's because rats are cute, warm and cuddly.

Ew, no they're not.

Well, OK, the domestic variety might be warm and potentially cuddly, but they're not cute. At least not to me.

Noone gives a fuck for experimentation on insects, or on cell cultures for that matter.

You'd be surprised.... although the folks protesting against experimentation insects do tend to be the most extreme of animal rights types.

have you ever read an ethics textbook

No. Do ethics textbooks exist at all? All I read on ethics was on philosophical textbooks.

Ethics textbooks do exist. Last time I was in my sister's medical library I browsed two of hers.

[Formless]

A brain is not a neural network as defined in computer science textbooks. It is a physical machine of action potentials floating in a modulating broth of fluid. That's not amenable to translation into logic functions.

While written in a professional and convincing style with the correct buzzwords (to the contrary of your street-dirt ramblings), the bolded sentence is pretty idiotic. Pretty goddamn fucking unbelieavably idiotic.

As opposed to your smiley ridden, smug, and horribly formatted posts. Right, mine are street dirt, yours are poetic and Kingly.

I like how you keep trying to paint me as an idiot while at the same time trying to incorporate my ideas into your approach despite what you originally proposed:

You know, to learn about how something works you must try to fuck some specific part up and see what happens, but doing it on actual human brains is uhhhhh... immoral.
Waiting for bad things to casually happen on people and studying them is very inefficient, and chimps are just related to us, but not the same.

Which is a far cry from:

Not exactly, I said we could make a model that mimicks the workings of a human brain and shape its "growth" so that at the end the reactions are within human variability. It's not anywhere near the same as predicting the outcome of all the interactions, although the results are similar.
It grows as we gather more info about how the human brain works, by making theories about why the model doesn't work as a human. It isn't a complex concept to understand, I think.

Regardless of how the brain works (and of our limited understanding of it), it follows specific and unchanging (if pretty complex) rules just like any other natural phenomenon.

Nice no-limits fallacy you got there.

I'm not saying that it is easy (it is indeed mindboggingly complex, but possible).

How do you know its possible? We're talking about a bio-chemical machine that is to this day more complex than any computer we've ever built.

That [having to simulate the environment the brain exists and grows within] adds to the total number of things you have to simulate, and brings in the possibility that there are unobserved variables you didn't simulate

Yes! That's exactly the point!
Seeing where the model doesn't react like a human brain will allow you to detect that there are variables that may have been left out. And so you can start looking for them.
How can you make make experiments to look for variables you don't even knew existed?
Divination? Illumination? SMS from Flying Spaghetti Monster?

Maybe that's the point of all those psychology experiments, twin studies, clinical studies, and natural observations of humans that have been conducted since these sciences began? Seriously, your arrogance is astounding. Its like you honestly think all the research that has been done up till now must be worthless simply because it wasn't high tech enough for you.

Do you have a reading comprehension problem? I already said that there is no "ideal human" you can model, shitwit.

You should not talk of things you clearly don't understand.

Enough with the exaggerated chest beating. It only makes you look like a blowhard when nearly half the points I've made are things you couldn't actually find a problem with in principle.

The "average" of population studies is usually the "ideal human" most other medical science experiments assume they are dealing with, and the "ideal target" of the testing of most medicines. It's the reason why not all medicines and procedures work well on all people and there are different medicines (not just different brands, that's another story) to cure the same illness.

Tell me, if there is an "ideal human" that you can model for the purposes of psychology/neuroscience, how come there are entire psychological disorders that appear in some cultures and not others?

This is just the kind of reasoning that psychology/neuroscience doesn't need. Whenever people have tried to fit groups of humans into classification schemes of how much they resemble the "ideal" human, I don't think I have to tell you it never ends well. Even when it is well meaning, like in medicine, there can be problems. Different ethnic groups have different rates of diseases, particularly of course genetic diseases, but also things like lactose intolerance and diabetes. The facts of demographics just don't lend themselves to such an approach.

To tie this in with your proposal, what you consider a normal human response depends on what culture your simulant considers itself a part of. Say you are testing to see how the simulant copes with stress. Would you consider shows of emotion to be a normal human response? What if it was socialized into an east asian culture like Japan where public shows of emotion are generally frowned upon?

[someone_else]

Right, mine are street dirt, yours are poetic and Kingly.

And you're envious, admit it.

I like how you keep trying to paint me as an idiot while at the same time trying to incorporate my ideas into your approach despite what you originally proposed

Well, that was what a layperson would think of this kind of experiments, and you tend to point out things that I actually took for granted (and quite a few completely idiotic things among them).

Nice no-limits fallacy you got there.

It's called scientific approach. Any natural phenomenon follows some rules that you try to find out to understand it. So far we found quite a bit of rules. I don't see why brain has to be different.

How do you know its possible? We're talking about a bio-chemical machine that is to this day more complex than any computer we've ever built.

Just as the Sun (or any star for that matter) is more efficient than any fusion reactor we ever built. How that is a good reason to declare something impossible? How you dare to declare something impossible just because it is just mindboggingly complex?

Maybe that's the point of all those psychology experiments, twin studies, clinical studies, and natural observations of humans that have been conducted since these sciences began?

This is another way to do it, that also allows faster theory-testing. With a model, you have the population studies results at the beginning of your research, and you work to make the model give the same results.

Its like you honestly think all the research that has been done up till now must be worthless simply because it wasn't high tech enough for you.

Where have I posted this crap?
A better way to do some research doesn't mean all research done before it with a different method is crap, and I'm certainly not claiming it.

Tell me, if there is an "ideal human" that you can model for the purposes of psychology/neuroscience

I think you have misunderstood. I said that the model (and most medical science experiments) aim at the "average" of the population studies.
Like assuming all humans weight around 80 kg or so even if it is a gross generalization.

To tie this in with your proposal, what you consider a normal human response depends on what culture your simulant considers itself a part of.

Not exactly. It depends from what response and how complex model we are talking about.
Models that lack a virtual "frontal lobe" won't give a shit about what culture they are from (parietal, temporal, occipital lobes, cerebellum and spinal cord just elaborate sensory feed and control muscle movement). And there is still a big deal to understand about how they work, so models may be useful.

When you start to model frontal lobes, then the culture becomes relevant indeed.

You seem to assume I claimed a gargantuan "One Size Fits All". That I wanted to replace any and all neurological experimentation with models and that with one single model you can do everything (never claimed that).
But the problem is only yours. Brain models are another tool in our toolbox, with some restrictions and some advantages.
I just pointed out how, given enough time, they may become something like the AI described by the OP.

Ew, no they're not.

[child shout]THEY ARE CUTE!!!!!![/child shout]
I was making a generalization, I don't expect all people in the world to love lab rats or guinea pigs of course.

The same for insects.

[Formless]

Well, that was what a layperson would think of this kind of experiments, and you tend to point out things that I actually took for granted (and quite a few completely idiotic things among them).

Care to name a few, you snide little asswipe? Or are you going to keep sniping about my intelligence when you continue to backpedal from your original proposal? Oh, not to mention that you don't realize you need a frontal lobe to display higher thought, but I'm getting ahead of myself.

It's called scientific approach.

No, its called "wanking", and its quite the opposite. That you can't tell the difference makes me think this conversation is going to end very quickly.

Just as the Sun (or any star for that matter) is more efficient than any fusion reactor we ever built. How that is a good reason to declare something impossible? How you dare to declare something impossible just because it is just mindboggingly complex?

Wow, I mean, wow. Making fusion reactors is one of the most balls-hard engineering problems ever conceived, and that is what you compare this to? This is just plain sad. Read the main websight, you pathetic waste of biomatter. See how many times you can find the words "limits" mentioned. Not everything that in theory can be solved in practice can. I can't believe on this of all websites that has to be said. Is this really the level of thought posters are allowed to get away with in SLAM now?

This is another way to do it, that also allows faster theory-testing. With a model, you have the population studies results at the beginning of your research, and you work to make the model give the same results.

Which again, you dismissed in your original post. But please, continue to backpedal. That's always amusing to watch.

Where have I posted this crap?
A better way to do some research doesn't mean all research done before it with a different method is crap, and I'm certainly not claiming it.

Here is a series of statements by you, in this thread. I'll even add emphasis to make it easier for you to follow along, which I should add has the advantage of making it easier to read (seriously, you get a little too eager to use formatting tags where they aren't necessary, and it just looks like hell, to say nothing of all the emoticons). See if you can't find the pattern:

You know, to learn about how something works you must try to fuck some specific part up and see what happens, but doing it on actual human brains is uhhhhh... immoral.
Waiting for bad things to casually happen on people and studying them is very inefficient, and chimps are just related to us, but not the same. [editors note: this is probably the single most damning statement, not only because it was your first in this thread but for its sheer irreconcilability with your current statements]

You can place the machine in situations that you cannot recreate on humans due to either moral implications or time constraints (hell, if you have to wait 20 years [editors note: a figure you pulled out of your ass] for the result of a single experiment you're going quite slow).

While written in a professional and convincing style with the correct buzzwords (to the contrary of your street-dirt ramblings), the bolded sentence [about how neuroscientists think the brain works] is pretty idiotic. Pretty goddamn fucking unbelieavably idiotic.
Regardless of how the brain works (and of our limited understanding of it), it follows specific and unchanging [ed. note: never mind that thing called "natural selection" that changes them over time] (if pretty complex) rules just like any other natural phenomenon.

...

Seeing where the model doesn't react like a human brain will allow you to detect that there are variables that may have been left out. And so you can start looking for them.
How can you make make experiments to look for variables you don't even knew existed?
Divination? Illumination? SMS from Flying Spaghetti Monster? [editors note: as if psychologists aren't aware of this problem even though the entire "nature vs nurture" debate is all about rooting out hidden variables!]

...

You should not talk of things you clearly don't understand.
The "average" of population studies is usually the "ideal human" most other medical science experiments assume they are dealing with, and the "ideal target" of the testing of most medicines. It's the reason why not all medicines and procedures work well on all people and there are different medicines (not just different brands, that's another story) to cure the same illness.
There are always differences between theory and practice, but the more advanced your knowledge gets, the less these "unexplainable rare differences" there are. [editors note: even though the differences are fully explainable through simple lack of data and variation among people]

Combined, these statements make it painfully clear that you have little to no idea how social science/brain science research is conducted. You make bold proclamations about how "inefficient" psychological experiments are, even though 1. most psychological data is gathered through observational methods other than experimentation such as (increasingly nowadays) brain imaging and 2. many psychology (especially cognitive and social psychology) experiments such as those done on attention blindness and the Milgram experiment were quite simple and low tech in their construction compared to, say, the Double Slit experiment or high energy particle physics. This, combined with your repeated attempts to mock my intelligence while simultaneously claiming the methodological problems I put forth were always assumed as part of your proposal, suggests that you do in fact believe those methods used by social scientists are comparatively worthless when placed next to the high tech wank approach. Denying it doesn't change what you have already posted.

I think you have misunderstood. I said that the model (and most medical science experiments) aim at the "average" of the population studies.
Like assuming all humans weight around 80 kg or so even if it is a gross generalization.

Even in medicine we have epidemiology for a reason. You just plain don't know what you are talking about.

Models that lack a virtual "frontal lobe" won't give a shit about what culture they are from (parietal, temporal, occipital lobes, cerebellum and spinal cord just elaborate sensory feed and control muscle movement).

They will also be incapable of higher thought. Or any thought. Really, this is so ignorant of basic neurobiology it could have been written by avianmosquito.

[Broomstick]

If I may interrupt this confrontation....

This article from the New York Times would seem to be at least a little relevant to our topic.

When five television studios became entangled in a Justice Department antitrust lawsuit against CBS, the cost was immense. As part of the obscure task of “discovery” — providing documents relevant to a lawsuit — the studios examined six million documents at a cost of more than $2.2 million, much of it to pay for a platoon of lawyers and paralegals who worked for months at high hourly rates.

“It’s a means of showing who leaked information, who’s influential,” said Elizabeth Charnock, founder of Cataphora, an information-sifting company.

But that was in 1978. Now, thanks to advances in artificial intelligence, “e-discovery” software can analyze documents in a fraction of the time for a fraction of the cost. In January, for example, Blackstone Discovery of Palo Alto, Calif., helped analyze 1.5 million documents for less than $100,000.

Some programs go beyond just finding documents with relevant terms at computer speeds. They can extract relevant concepts — like documents relevant to social protest in the Middle East — even in the absence of specific terms, and deduce patterns of behavior that would have eluded lawyers examining millions of documents.

“From a legal staffing viewpoint, it means that a lot of people who used to be allocated to conduct document review are no longer able to be billed out,” said Bill Herr, who as a lawyer at a major chemical company used to muster auditoriums of lawyers to read documents for weeks on end. “People get bored, people get headaches. Computers don’t.”

Computers are getting better at mimicking human reasoning — as viewers of “Jeopardy!” found out when they saw Watson beat its human opponents — and they are claiming work once done by people in high-paying professions. The number of computer chip designers, for example, has largely stagnated because powerful software programs replace the work once done by legions of logic designers and draftsmen.

Software is also making its way into tasks that were the exclusive province of human decision makers, like loan and mortgage officers and tax accountants.

These new forms of automation have renewed the debate over the economic consequences of technological progress.

David H. Autor, an economics professor at the Massachusetts Institute of Technology, says the United States economy is being “hollowed out.” New jobs, he says, are coming at the bottom of the economic pyramid, jobs in the middle are being lost to automation and outsourcing, and now job growth at the top is slowing because of automation.

“There is no reason to think that technology creates unemployment,” Professor Autor said. “Over the long run we find things for people to do. The harder question is, does changing technology always lead to better jobs? The answer is no.”

Automation of higher-level jobs is accelerating because of progress in computer science and linguistics. Only recently have researchers been able to test and refine algorithms on vast data samples, including a huge trove of e-mail from the Enron Corporation.

“The economic impact will be huge,” said Tom Mitchell, chairman of the machine learning department at Carnegie Mellon University in Pittsburgh. “We’re at the beginning of a 10-year period where we’re going to transition from computers that can’t understand language to a point where computers can understand quite a bit about language.”

Nowhere are these advances clearer than in the legal world.

E-discovery technologies generally fall into two broad categories that can be described as “linguistic” and “sociological.”

The most basic linguistic approach uses specific search words to find and sort relevant documents. More advanced programs filter documents through a large web of word and phrase definitions. A user who types “dog” will also find documents that mention “man’s best friend” and even the notion of a “walk.”

The sociological approach adds an inferential layer of analysis, mimicking the deductive powers of a human Sherlock Holmes. Engineers and linguists at Cataphora, an information-sifting company based in Silicon Valley, have their software mine documents for the activities and interactions of people — who did what when, and who talks to whom. The software seeks to visualize chains of events. It identifies discussions that might have taken place across e-mail, instant messages and telephone calls.

Then the computer pounces, so to speak, capturing “digital anomalies” that white-collar criminals often create in trying to hide their activities.

For example, it finds “call me” moments — those incidents when an employee decides to hide a particular action by having a private conversation. This usually involves switching media, perhaps from an e-mail conversation to instant messaging, telephone or even a face-to-face encounter.

“It doesn’t use keywords at all,” said Elizabeth Charnock, Cataphora’s founder. “But it’s a means of showing who leaked information, who’s influential in the organization or when a sensitive document like an S.E.C. filing is being edited an unusual number of times, or an unusual number of ways, by an unusual type or number of people.”

The Cataphora software can also recognize the sentiment in an e-mail message — whether a person is positive or negative, or what the company calls “loud talking” — unusual emphasis that might give hints that a document is about a stressful situation. The software can also detect subtle changes in the style of an e-mail communication.

A shift in an author’s e-mail style, from breezy to unusually formal, can raise a red flag about illegal activity.

“You tend to split a lot fewer infinitives when you think the F.B.I. might be reading your mail,” said Steve Roberts, Cataphora’s chief technology officer.

Another e-discovery company in Silicon Valley, Clearwell, has developed software that analyzes documents to find concepts rather than specific keywords, shortening the time required to locate relevant material in litigation.

Last year, Clearwell software was used by the law firm DLA Piper to search through a half-million documents under a court-imposed deadline of one week. Clearwell’s software analyzed and sorted 570,000 documents (each document can be many pages) in two days. The law firm used just one more day to identify 3,070 documents that were relevant to the court-ordered discovery motion.

Clearwell’s software uses language analysis and a visual way of representing general concepts found in documents to make it possible for a single lawyer to do work that might have once required hundreds.

“The catch here is information overload,” said Aaref A. Hilaly, Clearwell’s chief executive. “How do you zoom in to just the specific set of documents or facts that are relevant to the specific question? It’s not about search; it’s about sifting, and that’s what e-discovery software enables.”

For Neil Fraser, a lawyer at Milberg, a law firm based in New York, the Cataphora software provides a way to better understand the internal workings of corporations he sues, particularly when the real decision makers may be hidden from view.

He says the software allows him to find the ex-Pfc. Wintergreens in an organization — a reference to a lowly character in the novel “Catch-22” who wielded great power because he distributed mail to generals and was able to withhold it or dispatch it as he saw fit.

Such tools owe a debt to an unlikely, though appropriate, source: the electronic mail database known as the Enron Corpus.

In October 2003, Andrew McCallum, a computer scientist at the University of Massachusetts, Amherst, read that the federal government had a collection of more than five million messages from the prosecution of Enron.

He bought a copy of the database for $10,000 and made it freely available to academic and corporate researchers. Since then, it has become the foundation of a wealth of new science — and its value has endured, since privacy constraints usually keep large collections of e-mail out of reach. “It’s made a massive difference in the research community,” Dr. McCallum said.

The Enron Corpus has led to a better understanding of how language is used and how social networks function, and it has improved efforts to uncover social groups based on e-mail communication.

Now artificial intelligence software has taken a seat at the negotiating table.

Two months ago, Autonomy, an e-discovery company based in Britain, worked with defense lawyers in a lawsuit brought against a large oil and gas company. The plaintiffs showed up during a pretrial negotiation with a list of words intended to be used to help select documents for use in the lawsuit.

“The plaintiffs asked for 500 keywords to search on,” said Mike Sullivan, chief executive of Autonomy Protect, the company’s e-discovery division.

In response, he said, the defense lawyers used those words to analyze their own documents during the negotiations, and those results helped them bargain more effectively, Mr. Sullivan said.

Some specialists acknowledge that the technology has limits. “The documents that the process kicks out still have to be read by someone,” said Herbert L. Roitblat of OrcaTec, a consulting firm in Altanta.

Quantifying the employment impact of these new technologies is difficult. Mike Lynch, the founder of Autonomy, is convinced that “legal is a sector that will likely employ fewer, not more, people in the U.S. in the future.” He estimated that the shift from manual document discovery to e-discovery would lead to a manpower reduction in which one lawyer would suffice for work that once required 500 and that the newest generation of software, which can detect duplicates and find clusters of important documents on a particular topic, could cut the head count by another 50 percent.

The computers seem to be good at their new jobs. Mr. Herr, the former chemical company lawyer, used e-discovery software to reanalyze work his company’s lawyers did in the 1980s and ’90s. His human colleagues had been only 60 percent accurate, he found.

“Think about how much money had been spent to be slightly better than a coin toss,” he said.

The upshot is that AI - even such limited "smart systems" as we currently employ that are in no way sentient - are beginning to eliminate the mental equivalent of ditch-digging. As this form of labor can be tedious, unpleasent, and as uncomfortable as repetitive manual labor there is definitely some good in this, however, it does mean a whole bunch of people now have to find other work. That makes for a difficult and uncomfortable transition period. Between bias against those in middle age needing a new career, and no real structure for retraining people, for some this means diminished prospects for the rest of their lives. So what may be good on a societal level can be a disaster on a personal level. And that's why there is always some resistance to automation: the very real fear that one will be dumped on the career trash heap as obsolete and too old to do something new. If all those people being displaced were assured of a new career of at least the same earnings and other benefits as their old one I don't think you'd see as much resistance, if any.

Really, this has been a problem since the start of the industrial age, and we seem not much better at solving it than we did 100 or 200 years ago. Perhaps some resources should be devoted to this problem, as it would remove one of the obstacles to acceptance of AI and its cousins. Not to mention being the right thing to do on a humane level.

[Sarevok]

I hear automated checkouts are becoming starting to become common in shops, gas station etc in US. Supposedly they are now cutting into retail jobs that only people could do before. Is there any truth to this ?

[Lagmonster]

I hear automated checkouts are becoming starting to become common in shops, gas station etc in US. Supposedly they are now cutting into retail jobs that only people could do before. Is there any truth to this ?

They have them at most grocery stores and hardware stores here in Ottawa, all of the gas stations bar a few, and none of the shops. In all cases, they reduce the number of employees - at the discretion of the employer. If a customer doesn't want to check and bag his own groceries, he goes to the line and has a teenager do it. If a customer prefers to do things themselves, they will. I couldn't tell you which will win out.

In the rich part of Ottawa, there's a singular gas station that demands attention - the Island Park station. Its storefront is always clean (even in winter, they wash the walls and windows), the lights are bright, the grounds are swept and sprayed clean of dirt and oil. When you pull in, FOUR people run to the car. One checks your oil, one cleans the outside methodically - while smiling. A chipper girl with a bright smile in clean coveralls asks you how she can help and makes flirty small talk with you while a boy pumps premium gas into your tank (did I want regular? Possibly, but I got the feeling they would have reacted with horror if I'd asked). One of the boys says he notices your wipers could stand to be replaced, it'll only take a minute, would you be interested? You sit there, dumbfounded, wondering what the hell you're going to tip these people with. In a minute, my car is clean, full of strangely overcompensating gas, my oil has been checked, and I have new wiper blades. I've been intimidated into coughing up a hundred bucks plus tips just by the sideways glances of the BMW owners staring at my Hyundai. That's the standard of the business, and the rich people love it.

Or you drive up the road to a gas station erected in 1970 and built of what appears to be several aluminum canoes, and each pump has an automated do-it-yourself swipe card slot for payment and a pump-it-yourself policy. A dirty squeegee rests nearby in a bucket of something blue. The ground is covered in oil, salt, and grime. From inside a tiny, poorly-ventilated little kiosk, a single guy picks his nose and reads a book, glancing occasionally at a monitor but not paid to stop people if they decide to pump-and-dash, just report them and submit the video tapes of the license plate and theft to the police. He's technically there to help, but only if the pay-it-yourself pumps break, at which point he tapes a hand-written Out of Order sign on it and goes back to his book. You fill up your car, take a receipt, and drive off in half the time it took the pep squad over at the Island Park station. It costs you $40.

I have no idea what's better, but I do know that businesses are willing to try just about anything that sounds like a good idea.

[Broomstick]

I hear automated checkouts are becoming starting to become common in shops, gas station etc in US. Supposedly they are now cutting into retail jobs that only people could do before. Is there any truth to this ?

Around here, I'd say about half the grocery and DIY shops have self-serve checkout of this sort, and the number of such lanes seems to be expanding (they do retain human cashiers, but reduced in numbers)

Other stores refuse to install them.

So common, yes, but not universal. And possibly never universal, because there will always be people who want to go to a human and are even willing to pay a premium for it. Remember, there are still a few buggy whip makers out there, and even some people who still know how to knap stone tools out of flint rocks. Almost nothing vanishes completely, it's just becomes extremely rare.

[Broomstick]

My local Meijer's has one person "babysitting" 4 to 6 of those self serve checkouts, as opposed to having 3-5 additional cashiers. Yes, that does make fiscal sense. Unless every self-serve customer is 4 times slower than going through a lane with a cashier it represents a net gain for the corporation.

[Simon_Jester]

Only a few years ago, I happened to stop at a station up in Hagerstown, Maryland that still had a gas station attendant. I don't remember any of the details of exactly where and I didn't take a picture or anything... though I was sorely tempted to; for all I know, that's the last station in Maryland that has one aside from ridiculously high-end places like the one Lagomonster mentioned.

Automated checkouts leave a good deal to be desired, and probably will for the foreseeable future. They save labor costs if you build enough of them, but honestly I'm not sure a lot of grocery stores have enough physical space to put that many automated checkouts in. Not enough to handle peak traffic- so it's cheaper, but the lines are going to pile up longer than they used to, and I don't know how the math works out on the costs of running the things versus the cost of keeping the cashiers in place.