Cyberarchy: Yea or Nea?

[kardashev.livejournal.com]

A lot of people fear the idea of an artificial intelligence taking over the world and making decisions for the human race or wiping us out. You know, like the AI super-intelligences in The Matrix or Terminator. In fact, this is probably where they get their fear from. They even seem to fear a friendly AI running things. And yes, I know that AI might never become reality but for the sake of arguments assume that it can happen and that by virtue of always knowing the answer to a question it will come to rule us. At least in the de facto sense and probably de jure, as well.

Seriously, why fear a friendly AI? It will be hardwired to act in our own best interests. Human politicians aren't hardwired in this way. You won't be able to bribe such an AI. Or tempt it with sex. Or play on non-existent prejudices, petty grudges, or deeply rooted hatreds. It will have only cold hard logic to guide it after we turn it on. If the issue is "We need to provide food, shelter, and medical care for everybody" it will give us a completely unbiased answer of whether it is possible and how best to accomplish this. It may not give us a utopia(Hell, it may very well tell us that utopia is a pipe dream) but I bet it will be a lot more effective than letting human politicians and bureaucrats run things.

I think the real reason people fear the idea of an AI takeover is that they hate being told when a dream is impossible to achieve. Or that their ideas are demonstrably wrong. It's sort of like how the Maoists liked to put people in jail for being educated.

But me? Hell yeah, point me to the Machine God that I may hear something accurate for a change.

Comments

[kardashev.livejournal.com]

lol

[skull-bearer.livejournal.com]

[kardashev.livejournal.com]

lmao!

Thanks. This made my morning.

[kayjayuu.livejournal.com]

And if it decides with cold, hard logic that you as a seventy-year old are less likely to benefit than several small babies in whatever survival circumstances Datamaster is spitting out its logical figures, you'll just walk right into the light, wontcha?

[kardashev.livejournal.com]

You're assuming it will favor collective best interests over individual ones.

I doubt it will favor either one.

[pastorlenny.livejournal.com]

Only people who have no direct experience with IT make claims about the accuracy thereof.

[telemann.livejournal.com]

Ray Kurzweil, the guy who invented flat bed scanners (initially to read documents for visually disabled people), and created a synthesizer when he was in his teen age years, and computer whiz kid literally, would disagree very vehemently.

[malasadas.livejournal.com]

[onefatmusicnerd.livejournal.com]

Best Response!

[allhatnocattle.livejournal.com]

Just click...

http://interaxon.ca/

[kardashev.livejournal.com]

Oh, now that is interesting.

[fizzyland.livejournal.com]

Are you so quick to forget the Butlerian Jihad?

[kardashev.livejournal.com]

Were the machines of the Butlerian Jihad programmed with Asimov's Three Laws?

[soliloquy76.livejournal.com]

[kardashev.livejournal.com]

Insufficiently and incorrectly programmed. cf. GIGO.

[jubelhunden.livejournal.com]

Why should I be scared of Angry Birds taking over the world? They'd protect us from pigs...

[xforge.livejournal.com]

Read Mary Shelley's Frankenstein and then read Asimov's I, Robot and then get back to us.

[xforge.livejournal.com]

Yeah and then watch 2001: A Space Odyssey 'cuz that'll help too. = )

[sandwichwarrior.livejournal.com]

Seriously, why fear a friendly AI?

I'm not sure "fear" is quite the right word but I'd say the answer lies in the same reason that (I assume) you do not want live with your parents for the rest of your life.

Children grow up and leave the nest and many people chafe at the idea of be ruled or told what to do even if those descisions may be in their best interest.

To quote CS Lewis

Of all tyrannies, a tyranny exercised for the good of its victims may be the most oppressive. It may be better to live under robber barons than under omnipotent moral busybodies. The robber baron’s cruelty may sometimes sleep, his cupidity may at some point be satiated; but those who torment us for our own good will torment us without end for they do so with the approval of their own conscience. They may be more likely to go to Heaven yet at the same time likelier to make a Hell of earth. Their very kindness stings with intolerable insult. To be ‘cured’ against one’s will and cured of states which we may not regard as disease is to be put on a level of those who have not yet reached the age of reason or those who never will; to be classed with infants, imbeciles, and domestic animals.

[kardashev.livejournal.com]

"Children grow up and leave the nest and many people chafe at the idea of be ruled or told what to do even if those descisions may be in their best interest."

You wouldn't have to be "told" though. Such an AI could say, "Are you sure you want to continue with your present course of action?"

Humans say: "Yes!"

Failure results. AI says, "See? Told ya' so."

Enough of that eventually wears down the most stubborn folks. Having AI isn't like having parents. It's closer to having a investment manager, a doctor, an economist, or any other expert on your side. Difference being that the AI is an expert on nearly everything and is rarely ever wrong. It's the difference between "Go to your room! You're grounded!" and "See? I told ya so!".

[montecristo.livejournal.com]

What Is Wrong With That Picture (pt 1 of 2)

Your "cyberarchy" artificial intelligence paradigm suffers from what Friedrich Hayek called "the pretense of knowledge." The entire idea is premised upon having access to information that is just not obtainable by a centralized system. Human centralized command and control systems do not work not only because human beings are not perfectable, but also because human society is a complex, non-linear, chaotic system demonstrating emergent behavior and sensitivity to initial conditions. Such systems are inherently unpredictable and uncontrollable.

Look at what your AI would need to do. This has come up before, and one of you on this forum replied to this objection of mine with an intelligent, well-written article on calculability. Coming from a computer science background, I found the article interesting, and very similar to other writings on the topic, but the problem with it was that it failed to address the underlying philosophical and economic reality of the problem. I'm not denying that such an approach is an attempt to solve A problem, rather that the approach is prima facie evidence that the problem is not being understood.

The very root of the problem presupposes an objective, universal definition for something conceived of as the "collective good" or that there even exists an "individual good" that is objective, universal, and definable for all human beings in all contexts. The problem is that this is analogous to claiming a "priviledged frame of reference" in the physical world where one does not exist. Why do people postulate, or even desire, "control" over others? It is because they wish to substitute the judgements, values, and actions of some individuals with other judgements, values and actions. The question then becomes one of defining what is the source of judgements, actions, and values. The answer is: the individual. There is No "Social Mind." Hobbes's Leviathan and Hegel's God-Walking-the-Earth are metaphors, abstractions which cannot be concretized. They exist only in the imaginations of individual human minds. This is not to claim that they do not exist at all; "society" exists, in the same since that "love" exists. It is just not an objective, concrete, thing, in and of itself which can exist independent of the entities which comprise it. You cannot substitute, however benevolently you may desire to do so, some objective valuations, judgements, and actions that are objective, universal and context-independent, for the actions of individual human beings, because such things do not exist at all. The only thing that can be accomplished is to substitute some individual human beings valuations, judgements, and actions for the freely formulated valuations, judgements, and actions of other individual human beings. This raises the question: WHOSE valuations, judgements, and actions are to be considered "more equal" than whose? It does you no good to talk of democracy at that point; democracy implicitly presumes individual sovereignty as the basis for political equality. To simplify the question of just exactly who gets to decide, it reduces ultimately to what Vladimir Lennin explained is the essence of politics: "Who, whom?"

[montecristo.livejournal.com]

What Is Wrong With That Picture (pt 2 of 2)

The Church-Turing Thesis (http://en.wikipedia.org/wiki/Church%E2%80%93Turing_thesis) has been raised in here before, as an objection to the idea that only a free market can "calculate" an optimum economic set of circumstances. Basically, the objector is making a point that centralized command and control is not theoretically impossible, because any problem which is calculable by a set of distributed calculators can be theoretically calculated by a machine capable of simulating the operation of the distributed calculators. Of course, this objection misses the fundamental consideration underlying the problem. What are we "calculating" and what is doing the calculation?

The problem facing the collectivists is that individuals come together into society to solve problems cooperatively and to improve their individual circumstances. The "good" of society is the good of the individuals comprising it. The individual "goods" are determined and calculated by those individuals. If a computer were to attempt to calculate my ultimate good, it would first have to "know" me. Essentially, it would have to be me. In other words, it would have to render decisions that I would render myself in order to effect my ultimate happiness. Who determines what my ultimate happiness is? I do; I am the only one who can, since valuation is imputed: it is subjective and context-dependent. Of course, it may be true, that IF such a machine could be invented, which could determine, calculate, and effect each individual's "ultimate happiness," then with a sufficiently large enough and powerful enough Turing Machine, () one could replace all of the individual "happiness calculators" with one, centralized, happiness-calculating, economy-simulating machine. Here is the clincher though: at that point, "humanity" is rendered entirely superfluous. If you start out with a means that consists of overriding the individual's free will from the outset, you've essentially demonstrated that he is superfluous. Why is his preservation or optimization worth anything at all? It isn't, and this is logically necessarilly the case, given the thought experiments' premises about the acceptable means.

Consider, as Locke did, the converse situation: when the individual's good is no longer served by participating in a particular society, it behooves him to remove himself from that society and return to "a state of nature" in order to improve his circumstances. There is no "duty" to die for the collective. The collective has no existence apart from the individuals that comprise it. There is no justification for human sacrifice, no matter how "democratically" the victims are selected. As John Stuart Mill put it:

"If all mankind minus one, were of one opinion, and only one person were of the contrary opinion, mankind would be no more justified in silencing that one person, than he, if he had the power, would be justified in silencing mankind."

[kardashev.livejournal.com]

So, this would make AI no better than a human political figure? If so, I doubt that. Human politicians run into similar problems to what you describe, mainly that they cannot know every individual's "good". However, unlike human politicians the AIs would be beyond far more accurate in the questions they can answer, beyond temptation, treason, or other forms of self-interest. So far, I'm still willing to roll the dice with AI.

[terminator44.livejournal.com]

Glory to the Omnissiah!

it'll be worse than gay marriage

[paedraggaidin.livejournal.com]

[ddstory.livejournal.com]

Kinky!

[yes-justice.livejournal.com]

It will have only cold hard logic to guide it after we turn it on.

That seems an arbitrary limitation to set on something we already defined as AI.

Anyway, wake me when they can consistently tell a cat from a dog in 2d.

[montecristo.livejournal.com]

So, this would make AI no better than a human political figure?

If nothing can logically replace the individual as the sovereign acting unit within the economy then the category "nothing" embraces both politicians and artificial intelligences. You have to remember, even if the job were given to a hypothetical AI, it's value criteria would have to be chosen by....particular human beings. The problem with "electing" an AI is the same one we face in electing politicians. It just introduces a level of indirection.

Consider the case where the police break into someone's home on the suspicion that illegal drug activity is going on inside. The acting officers claim that "our drug-sniffing dog smelled something outside the front door" as a basis for their "reasonable suspicion." To turn the case into a question of how accurate the drug-sniffing dog is at detecting drugs is entirely to miss the point; it is still the cops' discretion as to what the dog's behavior means or even whether or not that behavior was observed at all! Even if you were able to create an AI capable of "manning" the bureaucracies of government, it's criteria for action, decision, and prioritization would have to be programmed or otherwise determined by some human being(s)! Value is imputed by individual human minds. It does not exist as an objective, context-free entity in and of itself which may be "imported" into the algorythms of some hypothetical AI. Your hypothetical AI cannot escape being the instrumentality of some human agency.

[montecristo.livejournal.com]

What you are proposing was already the subject of a "darkly humorous" role playing game that originated in 1984. It was called Paranoia (http://en.wikipedia.org/wiki/Paranoia_rpg).

Your scenario, if it were even possible to implement it, would turn Karl Marx on his head and be a case of history repeating itself, first as farce, second as tragedy.

[gunslnger.livejournal.com]

I'm fine with the machines being in charge, as long as I'm the one who programmed them. I don't trust any other programmers to do it right.

[geezer-also.livejournal.com]

Well being as how, I personally wouldn't be able to program them at all, I would say I might trust you more than most of the people who post in this forum, but still not enough to let you do it ;)

Who would program them just might be a problem :D

[peristaltor.livejournal.com]

They're kinda already here. Do a little digging on "high frequency trading" and note that computers are spending our money on investments. Sometimes it works, but. . . .

Now extrapolate the various "fat fingered" (read: we're not sure why it happened) financial trades that nearly brought down whole markets a few times to stuff that really matters.

[underlankers.livejournal.com]

Whenever I hear people saying that AIs will be necessarily an improvement, I sometimes wonder if they've ever heard of software rot or considered the problems of expecting an alien type of intelligence, assuming Hollywood-style AIs would ever be invented, to act like humans? If the attempt succeeds then humankind may well create an intelligent lifeform as flawed as Man himself is, but with an alien basis of that intelligence. The AIs might not even necessarily *want* to replace humans but their means of acting may be so different to ours that we'd be unable to realize it and they unable ever to adequately express it.

[mikeyxw.livejournal.com]

I'm kind of partial to the governing getting the consent of the governed from time to time. I know that this is probably not the most efficient process in the short term, but there are some merits to this idea. I'm not sure how the AI would do this, it seems like it would be a one-time deal or that it would need to appeal to people in the same way that politicians do today. Either we lose our right to pick our leaders or we come up with an AI version of today's politicians, assuming artificial stupidity would not prove too difficult to master by our machine overloards. Someone would probably program them not to lie too, making things insufferable.

[anfalicious.livejournal.com]

The sooner the robot overlords take over from the meat puppets the better.

[sophia-sadek.livejournal.com]

The Maoists were not the only ones to persecute the educated. Let us not forget McCarthyism.

It is interesting that you brought up this topic. I have worked on human/machine interface projects that enhance the capacity of the human brain. Naturally, there are some weak-kneed people who fear the power of the highly connected class.