Can artificial intelligences suffer from mental illness?

No, it doesn't. That entire mental construction is something the human brain creates - every entity in it, including the difference between "exactly" and "approximately" - concepts that don't really make sense applied to concepts like waves and galaxies, or physical objects in general if we are getting technical.

We are now in the position of having to address the matter of what "is" is. Do we want to go there?

That's like claiming the computer implements the paper and pencil emulation we noted was theoretically possible. I'm not persuaded that viewpoint has fundamental, or even reliable, implications.
For example:

Why are you assuming the universe has to round off its computations, if any? Or that if it does, they have to be large enough errors to produce effects visible to us? That reads like a modern version of Zeno's Paradox - a good question for thought, for clarifying one's thinking, but the arrow does reach the target.

 

Log in or Sign up to hide all adverts.

It doesn't. That's exactly the point. TMs must round off computations but the universe doesn't. The accumulated rounding errors eventually become huge. That's chaos.

 

Log in or Sign up to hide all adverts.

Tegmark believes that it's all mathematical, in the sense that all things have their own specific value and in theory their behavior can be calculated, at least in a short range and time.

His words; "some people say that the universe has some mathematical properties, I say the universe has only mathematical properties. It's just that we haven't discovered all of them yet."

 

Log in or Sign up to hide all adverts.

I know we had this convo several days ago. Correct me if I'm wrong.

I posted some thoughts on Tegmark earlier. Bottom line: He's interesting whether he's right or wrong. In my opinion he's right on some things, wrong on others. Wrong on MUH as I understand it.

Here are my comments on Tegmark.

http://www.sciforums.com/threads/ca...om-mental-illness.157386/page-14#post-3500529

 

Hmmmm, I must have missed that. I recall him saying that "if" we were part of a computer game, we wouldn't know the difference, because a computer's algorithms are approximate copies of natural functions, albeit it limited because it is impossible to "know" all the values and behaviors of the universe at nano scales.

There you have it then.

I pondered the concept of everything having a value (potential) or function which theoretically can be mathematically symbolized and I have not been able to identify anything which does not have an intrinsic value of some sort. Can you think of one?

 

There are far more mathematical objects than there are names for. Is that what you're asking? I'm a little unclear on your post. For example there are uncountably many real numbers, but only countably many finite strings of symbols taken from an at most countable alphabet. So there are many more real numbers than we can ever have names for.

So when you ask for each mathematical function in the universe to be able to by symbolized, you're rejecting all the mathematical functions that can not be symbolized. You're in effect denying that un-symbolizable mathematical functions exist in the world.

This relates to the fact that there are far more real numbers than can be computed by a Turing machine. There are only countably many Turing machines. Turing's 1936 paper was in fact directly about this. He was analyzing the set of real numbers that can be computed by a machine or automated procedure.

Now the thing about these noncomputable reals are that they are essentially random strings of bits. There is no pattern, no computer program that could crank out that bitstring.

So in a sense, your question is the same underlying question that we've been considering. Is the universe mechanistic? Or does it contain essential elements of randomness?

If the universe can physically instantiate a noncomputable real number, that would falsify the claim that the universe is a TM. But to instantiate such a number means the universe must be able to physically encode an actual infinite amount of information -- literally, the entire bitstring. It can't compress the bitstring down to a finite-length algorithm.

So that's the nub of the issue. Whether the universe instantiates a noncomputable real number. That's just another way of asking whether the universe is an algorithm or not.

 

Not true, Tegmark identifies some 33 specific numbers and a handful of equations, which account for everything we can measure at macroscale.
How would you write 2 trillion, quadrillion, quadrillion, quadrillion. Simple: 2 x 10^54

Can you name one at macro scale?

That's why we are using equations. (i.e. shorthand for very large numbers)

If you add probabilism, you can go a long way.

I disagree with that, anything that has a value can be symbolized in some way. Moreover you are not talking macro here.

Why is it necessary to have one single algorithm? Computers already have several different algorithms for specific functions.

A slime mold has no brain at all, it's a cingle celled organism. Yet it can solve a maze or copy a highway system that took years for engineers to develop. But in fact it does employ mathematics albeit in rudimentary form.

https://en.wikipedia.org/wiki/Slime_mold

And this amazing presentation:
https://www.ted.com/talks/heather_barnett_what_humans_can_learn_from_semi_intelligent_slime_1

 

I prefer not to go back and forth about MUH at the moment since I am a little busy with other things and that's a big subject. So this hopefully will be my last post on Tegmark today.

That's one of his errors. There are 33 constants ("There are 57 communists in the State department!" as Joe McCarthy used to say) that account for everything we can measure. So those constants delineate the limits of our ability to measure. They say nothing about the world itself. The human-made, historically contingent THEORY has 33 control knobs. How many control knobs reality itself has, or whether it even has control knobs at all, is unknown.

The fact that some real numbers have names is irrelevant to the fact that most don't.

Mind.

The question was, what un-symbolizable mathematical functions might exist in the world. Mind is exactly it. It's a noncomputable real. It's a physical instantiation of actual infinity. It's the leap in physics that we need in order to solve the problem of mind.

Right. We use equations to model the results of our observations. The model only needs 33 constants and 31 flavors at Baskin-Robbins. [That's the slogan of an ice cream chain in the US].

What does that mean and what long way does it let us go? So God plays dice after all? Then nothing is certain. There's no determinism, God flips a coin. That your metaphysics?

You do have a point. Whether a noncomputable real number can be said to "have a value" is a pretty good question. There does happen to be a class of noncomputable numbers that can be given names; that is, they can be defined but not computed. Chaitin's Omega is one such. It's the probability that a random Turing machine halts. That's its definition. It can't possibly be computed, because computing it would solve the Halting problem, which is impossible.

https://en.wikipedia.org/wiki/Chaitin's_constant

But I'll stipulate that the question of whether an arbitrary noncomputable number can be said to have a value is a bit of a philosophical conundrum. They're very strange numbers. However it is undeniable that each noncomputable real does live at a particular, specific place on the number line. Whether that gives it a value, I can't say.

Turing showed there's a universal algorithm that's sufficient to compute any particular algorithm. This is the idea of the UTM, or universal Turing machine.

https://en.wikipedia.org/wiki/Universal_Turing_machine

The abstract idea of the UTM is the underlying idea of the modern general-purpose computer. You don't need one computer for word processing, another for watching cat videos, and still another to surf the web. A universal computer can compute anything that's computable. You have one in front of you right now as you read this.

Again, please forgive if I don't snap back right away.

 

The Universe is not sentient .It does not figure out anything

The stuff within the Universe arrange via PHYSICS

Please Register or Log in to view the hidden image!

 

Universe is not a computer

It computes nothing

Please Register or Log in to view the hidden image!

 

Physics has mathematical constructs

Please Register or Log in to view the hidden image!

 

Slime mild does not solve a maze. To solve something you should be detached from the problem and word it out .Slime mild reacts to trial and error

There is no problem in a maze

In the same way eyesight solved the problem of vision. There was no problem of vision.

Please Register or Log in to view the hidden image!

 

I agree, but "via physics" is a method of computation, is it not? It's just not a binary function.

 

Precisely, it deals with values and patterns (in the most generic sense)

But note that you stated that the universe arranges "via physics" and now you recognize that physics consists of mathematical constructs (inherent values). So if we combine both statements we arrive at;
"The universe arranges via mathematical constructs" (or patterns), which should, in principle, make it computable.

A similar computable function happens in inorganic chemistry, except it uses a different algorithm, chiral (left/right handed) connectivity, or EM algorithms which are based on electrical polarity.

The one thing which seems to present a real problem is in bio-chemistry, which apparently possesses some fundamental difference, which seems to be able to present itself in infinite forms from just a few hundred different bio-molecules.

. Have you ever been in a lifesize maze, which does not allow for visual orientation? How do you solve that problem other than by trial and error?

Except the slime mold does not have eyes or even a brain for that matter. Yet it seems to be present all over the world, obviously using a successful survival strategy.

If we used the example of filling the maze with slime mold as a substitute for external visual observation, the slime mold beats humans in finding the shortest route to the food, every time.

There seems to be a great scientific interest in this remakable brainless organism. I find that extraordinary and may possibly yield some new science inside the field of bio-chemistry itself.

It's behavior seems very similar to a hive-mind: "E pluribus unum"

 

You're taking my colloquial usage literally. Whenever we work with computers we say, "The computer thinks it should push the king's pawn one space forward," or, "The computer wants to balance the smart quotes in a Word document." We do not actually mean to say the computer thinks or wants. We casually anthropomorphize the computer ... knowing perfectly well that we're doing this. We're not making a metaphysical claim.

The point is that one way or another every particle in the universe ends up exactly where it's supposed to be at every moment.

There are three possibilities for how this happens:

* It's essentially random. There's no rhyme or reason to it. Any apparent order is either a man-made illusion, like seeing constellations in the sky; or else a momentary local coherence in an otherwise random universe.

* It operates according to laws; but these laws can not be perfectly implemented by a Turing machine. [It doesn't matter that you can always approximate the right answer with a TM. The universe does not put each particle in the approximately correct place, it puts it in the exact place]. In this case what the universe is doing may be called a computation. It determines what should happen according to rules. But it's a type of computation that we can't even imagine. It violates the Church-Turing thesis.

* The universe is a TM. In which case I'm wrong and Searle's wrong, The Chinese room understands Chinese. And we ourselves are no different in principle than a copy of Microsoft Word running on Windows 3.1 in 1995.

Those are your choices.

 

No

A body in space has no choice

It is going where it has to go

No choice in the matter. The game is fixed

It has mathmatical constructs in that you, as a sentinent entity, can use the maths to work out where (compute) the space body, will end up

But the body does not compute. Like a jigsaw puzzle. Drop the individual pieces down and you can compute where they will land. With the fix in they will form the picture when they fall

The inputs are constantly changing and new outputs are constantly being produced. Sometimes (learnt behaviour) a program already exists and is updated

Hampton Court Maze. Right hand on the bush, walk. Think with new mazes this method no longer works

However a maze is just a maze. It takes human thought to consider it a problem. Ignore it, so called problem, solved

Rapid adaptability to changing conditions. Nothing else required

If you have mold on you get rid of it quick before you become one of their changing conditions they rapidly adapt to

Good for them and their rapid adaptability

OR

Quid est sunt nobis faciens hic

Please Register or Log in to view the hidden image!

 

I'll go with this one but i will do some checking to confirm

Please Register or Log in to view the hidden image!

 

Sorry

Please Register or Log in to view the hidden image!

my literal 3 neurone brain Huey Dewey and Louie

Please Register or Log in to view the hidden image!

 

Will still stick with this one

Did a little digging. Lot of mathematics involved

Please Register or Log in to view the hidden image!

i don't do mathematics very well

If you can explain in words of two syllables with crayons it MIGHT help me understand

Please Register or Log in to view the hidden image!

 

No math required.

In the 1920's people were wondering whether there was an "effective procedure," whatever that might mean, for solving certain problems in number theory. In other words given a problem of a certain type, was there an algorithm (our modern word for what they were thinking of) that would terminate within a finite number of steps. That last requirement is a crucial part of the idea, since given a problem in number theory we can just try every possible integer if we are allowed infinitely many steps.

For example to see if there's an integer solution to the Fermat equation a^3 + b^3 = c^3, we could just try every possible combination of a, b, and c if we are allowed infinitely many steps. So the game is to try to:

* Define formally what we mean by an effective procedure; and

* Using that definition, determine whether there's an effective procedure to solve problems in number theory.

In 1936 Turing defined his idea of the Turing machine, showed that it's a reasonable definition of effective procedure, and showed that there were problems in number theory that could not be solved by any effective procedure. His work was strongly related to other work of the same era by Gödel, Church, Post, and others.

It was soon proven that the approaches to defining effective procedures given by Turing, Post, and Church were equivalent. They said the exact same thing in different ways.

So now the question becomes: Is there any kind of computation or effective procedure that's NOT a TM? Something that we would recognize as a computation, but that is not reducible to a TM?

The Church-Turing thesis says that there is not such a thing. It says in effect that any computation that can be done by a machine in the physical world is reducible to a TM.

In the 70 or 80 years since this thesis was proposed, no exception has been found. Quantum computers are reducible in principle to TMs, as are neural nets.

So when people say that the universe is a computation, the question is whether

a) It's reducible to a TM; or

b) It's a new type of computation that isn't reducible to a TM, thereby going beyond the Church-Turing thesis.

There's a third possibility:

c) The universe isn't a computation or machine at all. It's essentially random.

That's the question at issue. Because if the universe is a TM then so is the human mind. It's known that any TM can be implemented on any hardware. I can run a neural network on a supercomputer or I can execute it with pencil and paper and I'll get the same result. So if the universe and/or the mind are TMs, then they can be run in the wetware of the brain or in the digital logic of a computer or by a person using pencil and paper. You could indeed "upload" your mind to a digital computer. If a human could go crazy, so could a computer mind, because there would be no difference in principle between Microsoft Word and your mind. They're both reducible to Turing machines.

On the other hand if one claims (as I do) that the mind is NOT a TM, then I am claiming that whatever the mind is, it goes past the Church-Turing barrier to some new kind of computation; or else it's essentially random and does not work according to any laws or rules at all.

(ps) I'll add that it's in the realm of randomness that I think the answer lies. We can easily go past the limitations of TMs by allowing various forms of infinitary computations. There's a large literature on this and the related idea of hypercomputation, which people have probably heard of.

All these ideas remain theoretical, because the universe can not, as far as we know, instantiate an actual infinity.

If we get a revolution in physics that allows the universe to instantiate even a single noncomputable real, then infinitary computations become practical and we may have a better explanation of mind. The next revolution in physics will concern the infinite. We'll figure out what is the meaning of transfinite numbers and how they relate to what the universe does. Analogous to the way non-Euclidean geometry seemed nutty in the 1840's, and turned out to be the actual geometry of our world by 1915. Cantor's crazy numbers will find physical applicability, leading to a revolution in the theory of computation; which will give us radical new explanations for the nature of the mind and of the world.