ZOMBIE FORUMS

Ouch. That's kind of mean.

In my experience, if a computer does something it shouldn't be doing, you can tack it down to a programmer that didn't know what he was doing (Microsoft example alrady explained). The machne is predictable. The human being behind it is not. The computer is victim, just trying to do what it was told to do by crazy people, so it looks like the computer is crazy.

See, easily understandable.

You're just shifting the blame for the craziness, not eliminating it.

What does it matter if the AI was programmed by a crazy or sloppy programmer, evolved from some simple algorithms or had its genesis from the software of a toaster?

A sufficiently complex state machine is no different from a human brain... or shall we say... the human brain is a complex state machine. If you can predict the behavior of a sufficiently complex AI, then you can predict the same for a human.

Like when Jon and Sarine are gonna screw again?

I'd give it 30 pages.

Which is to say, you can't. The key element that distinguishes a human brain from a computer is its inherent unpredictability. Instead of following a simple, ergonomic path to initiate an action or react to a situation, the human brain is constantly churning a finite but large number of subroutines that trigger when a roughly correct set of stimuli occur.

This makes human reactions unpredictable, but also highly adaptable. We are constantly overthinking our actions and our responses to the environment and can react to almost any random change.

Which is why computers both seem and are worse at responding to emergency situations than humans. With a computer, you have to think of every problem that might FUBAR the situation, or invent an algorithim that can respond to a range of problems. Per Murphy's Law, something you did not predict or could not efficiently allocate processing for will occur and all the airliners will try to land on the same luggage cart at the same time. The human brain can react to all the things that are screwy about that last sentence and raise an alarm.

So machine programs can be analyzed and repaired to an absolute, if someone puts the effort in. Human programming will never be completely correct or predictable. By and large, though, it will not mistake a nuclear missile launch console for a computer game.

What was it we were talking about?

You guys fucken derailed this thread sooooooooo bad

Well, sometimes it's a programmer who doesn't know what he's doing, and sometimes it's a programmer who doesn't know what he's done. Not all typos result in something that won't compile and run, unfortunately. (And sometimes a bug hunt feels more like a snipe hunt...)

If a machine starts doing something I didn't tell it to do, I'm gonna take an axe to it.

OTOH, I think there are some higher reaches of fractal and genetic programming which don't readily conform to predictability; there's enough pseudo-randomness going on to make the answers surprising (I speak with no experience of programming in said disciplines; I surmise). But most of us don't work in such rarefied strata.

Perl just makes the job of understanding what's going on that much harder. No, I do not want a flame-war.

The human brain may be a complex state machine, but the human mind including all its myriad levels, is not.



Nix, by and large, that which distinguishes a human mind is its relative predictability. It is the unpredictable situations that make for 'unpredictable' reactions. Noting that no situation occurs ahistorically, previous occurances would add to the situational matrix and the alter or reinforce human reaction making for 'unpredictability'.

The human mind is a gestalt of relatively stable schemas, you know how hard it is to make someone change their habits, of action and of thought, although it is certainly possible to do so. Indeed, if the human mind were inherently unpredictable, there would be no stable personalities. The human mind is inherently stable but with the ability to learn and change.


;)



Something about magic elven manacles and a half-elf bondage I think.

*zaps Kian with an Obelisk of Light*

*burns Kian's remains with a flame tank*

I'm not sure the human brain *is* a state machine. After all, its functions are the instantaneous but never-ending transfer of electrical impulses between nerve cells. A particular synapse is only meaningful if it's followed up with more synapses. With computer, OTOH, (and this is more obviously of ferrite core memory machines than VRAM micros), the state at a given moment is the purpose, and the transfer of electrical signals within the system is done only for the purpose changing the present state to something else.

Compare the biological and physical definitions of work.

1) The human brain can have only a finite number of states. As it isn't infinitely large, and each neuron can only have a finite (non-binary sure, but still finite) number of states.
2) The speed at which a neuron fires, (the effective communication speed) is only around 200 mph. So you can't build a brain very large.
3) Just because we don't understand how it works, doesn't mean it isn't deterministic, which is all that is required.

.'. A human brain is a finite state machine. It follows the laws of other Turing Machines. Before you say "OMG that doesn't work, I can tell if a computer program will exit, that's easy." Try this one:



This "program" will run until it encounters a twin prime and then it will exit, because we don't know if there's an infinite amount of twin primes, no amount of simple inspection will tell us whether or not this program halts.

The brain is nothing but a glorified computer, but don't worry, I won't tell anybody that we're not special and we don't somehow break the laws of mathematics.

I agree that the brain maybe a "glorified computer". But that can't be the only thing that's going on up there. There's also the little matter of sentient behaviour, the ghost-in-the-shell, self-determination. Even given the Chinese box thought experiment, and other tests for being human. To state the obvious, my Mac doesn't go down the pub, start developing tastes in wierd music, webcomics etc. It does what it's told, and that's the way I likes it. So what else is going? I think that rests in the realm of biology not maths.

I love it when mathematicians, philosophers, and computer programers start waxing pompous as if they know a fucking thing about how the human mind works.

Hmmm . . . no, "relative" predictability doesn't cut it. Either you are talking about determinism, or about probabilities. A human brain has a range of responses possible to any situation, with no finite limits on either either end of the interaction. In general, you can assign probabilities to human reactions, but you have to set arbitrary bounds to a situation to limit human reaction to a finite number. A computer has a fixed and limited number of responses to a fixed and limited number of stimuli. It is non-functional otherwise. Animals fall somewhere between computer programs and the human mind in degree of determinism, which directly implies that the human mind is a just another biological engine, and there is no separate conscious entity "inside" the brain. Which is to say, you could build a sentient computer, with consciousness and a "ghost," but no one is even vaguely close to doing so.


True, but we are not talking about determinism at that level, but at the most basic level of function. At the high programming level you describe, you are correct. However, no one has every made a machine that functions in this way.


Yup. Any machine or program can err. That is not the topic here.


Infinity is not a requirement. All that is required for non-determinism is a scale of connectivity too high for deterministic analysis. You don't even need something as complex as a brain for that.


The physical limits of the computing machine are a minor matter. Connectivity is what counts. Adaptability is the advantage the mind has over the machine. The mind, as I said before, is incapable of handling any stimulis and response situation in the most energy efficient way possible. That tangle of active programming is always there, observing the situation and applying itself as needed. A computer has to be able to reproduce that , that chaos of connections existing independently of current need, to be considered sentient.


As I noted above per the climb in complexity from computers through animals to humans, the brain probably is determistic in theory, but it cannot be in practice, at least at our level of technolgy, and probably not by any means available in the near future. To make a computing device adaptable to any problem, our best solutions are still sex, reproduction, and two decades or so of program growth.


We don't break the laws of statistics, but the more deterministic laws cannot provide answers for a human brain the way they do for a computer. Not a problem, really. Physicists had to handle that issue back when they invented quantum mechanics. Pissed off Einstein no end, but he had to live with it.

hrmm.. just checking you are talking about behaviour in the macro-scale world with me but micro-scale biological functioning of the neuron below, yes?

In this case, yes, the functioning of a neuron of the brain is deterministic. No two ways about that. We're in agreement methinks. :)



In the macro-world, if you know all the states of a person's brain's functions, including on stuff that we now not know, and on the exact effects of all environmental factors, it should be theoretically possible to determine his next action. Which is technically impossible for any one or any organisation to do, since it'll require knowledge of the entire causal universe of every atom and every field of energy at that instantaneous moment.

So although we at our present level of technical and theoretical understanding cannot determine the human mind at the macro level, it is theoretically possible. We can still deal with probabilities though, if a guy is trying to kill another guy and knowing a few other factors, we can generally identify the most probably courses of action that will be taken. Indeed, it is possible to increase the probability of certain actions to be taken by the use of certain actions in a experimental study.

Killjoy, we know enough to discuss it, if we can only dicsuss something when we know exactly how it works, we'd never have figured out how to use the sling!

Well, there you go. The human brain always exits, sooner or later.

BTW, you've got it backwards. A Turing machine is a type of finite-state machine, not the other way around.