Wait, can free will exist?

For the discussion of the sciences. Physics problems, chemistry equations, biology weirdness, it all goes here.

Moderators: gmalivuk, Moderators General, Prelates

User avatar
Malconstant
Posts: 258
Joined: Wed Aug 24, 2011 1:44 am UTC

Re: Wait, can free will exist?

Postby Malconstant » Wed Oct 26, 2011 5:16 pm UTC

Anaphase wrote:From Turing's original paper:

Spoiler:
Interrogator In the first line of your sonnet which reads 'Shall I compare thee to a summer's day', would not 'a spring day' do as well or better?

Computer It wouldn't scan.

Interrogator How about 'a winter's day'? That would scan all right.

Computer Yes, but nobody wants to be compared to a winter's day.

Interrogator Would you say Mr. Pickwick reminded you of Christmas?

Computer In a way.

Interrogator Yet Christmas is a winter's day, and I do not think Mr Pickwick would mind the comparison

Computer I don't think you're serious. By a winter's day one means a typical winter's day, rather than a special one like Christmas.

There are no chatbots that can hold up under that sort of grilling.


Hogwash. I'll elaborate on this in a bit.

Anaphase wrote:
Malconstant wrote:The test is just to have a text chat with something while actively trying to figure out whether or not the thing you're talking with is a bot or if it's actually intelligent life.

The key word being 'actively'. Idly chatting while passively trying to determine what's on the other end is not the same.

Yeah, notice how I included that in my definition. I'm aware of this.

Anaphase wrote:
Malconstant wrote:It doesn't make any sense to imagine a computer that has universally passed the Turing test. Because if it existed and you were aware of it, then you would know that it was a computer, and thus if you were to text chat with it it would automatically fail the test for you.

The test is explicitly blinded, so I'm not sure what you're driving at.

You seem to be thinking that the Turing test must be administered by a person of sufficient intelligence. I'm saying that any old person will do the trick on the basis that there should be a fundamentally unbridgeable gap between any person and any look-up table. Surely you'd agree that the bar for passing a Turing test is different depending on who is the interrogator. If I made a chat bot that fooled a 6 year old, would you then be convinced that a computer has passed the Turing test?

Anaphase wrote: I'm saying no-one would seriously consider that a computer that passes such a test was using look-up tables. You do realize just how big such a table would have to be?

depends heavily on the length of conversation. The point that I'm making is theoretical possibility, and also knowing that humans aren't the best at not getting tricked. It's ridiculous to believe that it's impossible to write up a sophisticated enough chat box to fool someone who is actively trying to tell whether or not it's a bot.

And it's absolutely provable that such a look-up table chat box can be constructed on a theoretical basis, given an arbitrarily long (but surely finite) amount of time to write it. So what are you trying to argue again?

Oh right, christ the Turing test is obviously unrelated to free will. And my example of proving the theoretical (albeit difficult and impractical) ability to write such a look-up table computer should unequivocally demonstrate this. Please stop trying to argue against that point, this is what we call a proof in thought experiments.
Technical Ben wrote:PS, doogly, way to miss the point.

User avatar
LaserGuy
Posts: 4585
Joined: Thu Jan 15, 2009 5:33 pm UTC

Re: Wait, can free will exist?

Postby LaserGuy » Wed Oct 26, 2011 5:28 pm UTC

tomandlu wrote:
LaserGuy wrote:"Free will" works in much the same way. It is often a useful approximation for us to assume that we have control over our decision-making process, because it appears to us that we do. Scientific evidence, however, points to the contrary notion that, in fact, our "free will" and decision-making processes are determined entirely by our brain chemistry. In practice, it may be impossible to actually perform the calculations to figure out what the decision-making process is actually doing, but that doesn't mean that it isn't possible in principle to do so


I'm unconvinced that free-will has to mean "not predictable". The dice analogy is inadequate because, even if on some fundamental level, the dice roll was genuinely random, it still wouldn't follow that the dice had free will. Free will, if nothing else, requires conciousness. Free will requires self-awareness - it is to act with a perception of the assumed outcome of the action. Randomness is a distraction.


I'm not arguing that the dice have free will. That would be ludicrous. I'm arguing that the dice are genuinely deterministic, but under most conditions, there is nothing wrong with treating them as though they are random. Likewise, while it may be useful under some circumstances to describe human actions in terms of free will, that doesn't mean that the process that governs those actions is actually free will. The analogy is describing the difference between our perception of the system, and how the system actually works.

Anaphase
Posts: 33
Joined: Fri Oct 21, 2011 7:20 pm UTC

Re: Wait, can free will exist?

Postby Anaphase » Wed Oct 26, 2011 9:00 pm UTC

Malconstant wrote:
Anaphase wrote:
Malconstant wrote:It doesn't make any sense to imagine a computer that has universally passed the Turing test. Because if it existed and you were aware of it, then you would know that it was a computer, and thus if you were to text chat with it it would automatically fail the test for you.

The test is explicitly blinded, so I'm not sure what you're driving at.

You seem to be thinking that the Turing test must be administered by a person of sufficient intelligence. I'm saying that any old person will do the trick on the basis that there should be a fundamentally unbridgeable gap between any person and any look-up table. Surely you'd agree that the bar for passing a Turing test is different depending on who is the interrogator. If I made a chat bot that fooled a 6 year old, would you then be convinced that a computer has passed the Turing test?

I'm not sure what any of that has to do the the test being blind, I'll treat the remainder below...
Malconstant wrote:
Anaphase wrote:I'm saying no-one would seriously consider that a computer that passes such a test was using look-up tables. You do realize just how big such a table would have to be?
depends heavily on the length of conversation.

To put it mildly. If each word in a well-formed sentence represents on average five bits of entropy, then there are 250 ~ 1015 possible ten-word sentences. Even if we neglect longer sentences, each question-response pair added multiplies the size of the table by a factor of ~1015.
Malconstant wrote:The point that I'm making is theoretical possibility, and also knowing that humans aren't the best at not getting tricked. It's ridiculous to believe that it's impossible to write up a sophisticated enough chat box to fool someone who is actively trying to tell whether or not it's a bot.

And it's absolutely provable that such a look-up table chat box can be constructed on a theoretical basis, given an arbitrarily long (but surely finite) amount of time to write it. So what are you trying to argue again?

That conflates two radically different ideas, the mathematically possible but physically absurd exhaustive lookup table, and a simple chatbox using a grab-bag of responses. The ELT is as implausible as alien tricksters or diabolical TV producers who look like Ed Harris, and (at least to a non-Scientologist) is impossible to compile within the lifetime of the Universe.

That a simple chatbox, OTOH, might fool a naive cross-examiner isn't really relevant. The use of natural language isn't the point, a parrot can duplicate human speech, the purpose of the cross-examination is to probe the semantic depth of the interrogatee's understanding of the world. Provided the human subjects are not themselves intentionally evasive, it's not at all a simple matter to write a program that can fool even a moderately savvy judge. To date there are no programs, of any sort, that can maintain an appearance of understanding under those circumstances. There have been a few "limited Turing tests" performed as a stunt that narrowly restricted the judges' range of questions and required the human subjects not to give away any "tells", but that's about as meaningful as a "limited invisibility test" that requires the judge to face the opposite direction.

Malconstant wrote:Oh right, christ the Turing test is obviously unrelated to free will. And my example of proving the theoretical (albeit difficult and impractical) ability to write such a look-up table computer should unequivocally demonstrate this.

The point was that if a system clearly lacking "free will" were capable of behaving indistinguishably from one that supposedly has it, it would call into question whether anything that could possibly be pegged as "free will" could be worthy of the name.

User avatar
Malconstant
Posts: 258
Joined: Wed Aug 24, 2011 1:44 am UTC

Re: Wait, can free will exist?

Postby Malconstant » Wed Oct 26, 2011 9:59 pm UTC

brains in a vat have nothing to do with this. Why are you so insistent on conflating the relative probabilities of this? Your sense of probabilities is incredibly meaningless. The look-up table computer which can beat the Turing test is clearly possible. You shouldn't have to actually appeal to a mathematical argument to show this. But if you need that to convince you of your absurdly high standards than so be it. It's a thing that is possible, and if it's a thing that is possible, and you require its negation to be your free will test, there goes your free will. the end. that's what a proof means.
Technical Ben wrote:PS, doogly, way to miss the point.

User avatar
LaserGuy
Posts: 4585
Joined: Thu Jan 15, 2009 5:33 pm UTC

Re: Wait, can free will exist?

Postby LaserGuy » Thu Oct 27, 2011 3:50 am UTC

Anaphase wrote:That a simple chatbox, OTOH, might fool a naive cross-examiner isn't really relevant. The use of natural language isn't the point, a parrot can duplicate human speech, the purpose of the cross-examination is to probe the semantic depth of the interrogatee's understanding of the world. Provided the human subjects are not themselves intentionally evasive, it's not at all a simple matter to write a program that can fool even a moderately savvy judge. To date there are no programs, of any sort, that can maintain an appearance of understanding under those circumstances. There have been a few "limited Turing tests" performed as a stunt that narrowly restricted the judges' range of questions and required the human subjects not to give away any "tells", but that's about as meaningful as a "limited invisibility test" that requires the judge to face the opposite direction.


FWIW, in the 2008 Loebner Prize (one of the only formal Turing competitions in the world), the top chatbot managed to convince 3/12 of the judges that it was human (or at least more likely to be human that a parallel human who was also being interrogated) in a 5 minute unrestricted conversation.

Technical Ben
Posts: 2986
Joined: Tue May 27, 2008 10:42 pm UTC

Re: Wait, can free will exist?

Postby Technical Ben » Fri Oct 28, 2011 8:18 am UTC

gmalivuk wrote:
Technical Ben wrote:So your family would count as being Santa.
Absolutely not. At least, not if Santa is defined as being a man who lives at the North Pole and/or has a flying sleigh and/or delivers presents to everyone on the same night. My parents are exactly zero of those things.
Spoiler:
Is it not the results that matter?
But the result of getting presents does not and cannot differentiate between Santa existing and Santa not existing. So while it matters in the sense that I enjoy presents as much as the next kid, it doesn't matter in the sense of telling me Santa exists.

the result is "we still get the observation of free will". We still get the result we expect from free will no matter what the mechanism is.
But it's also the result we expect from a complete lack of free will. As such, it's still a shitty experiment.

way to miss the point.
It's hilarious to me that you would have the gall to say this to anyone else in this discussion, since it's you who continues to doggedly miss pretty much all of the points everyone else has been making all along
.


But you got presents right? Do you conclude presents appear "magically" or that they have an underlying mechanism? Your given a possible mechanism. At that age, it was impossible to test for. Later you got the correct mechanism, it was your parents not Santa.

You said free will is impossible to test for (in all possible universes). So, with the observation that "I" or "self" makes the choice, I make a model based on my observation. Not that "Santa" exists, but that "a method of choice" exists. As other objects are observable or testable I define them as "determined" or "random". But not being able to observe choice or free will it becomes "undefinable (free) method of choice (will)". I don't see it being able to be defined at any level, fundamentally or by physics.

I see the problem as, that as soon as the system becomes defined or definite, it can change. It self references, or self modifies. So it would not allow you to define a future state. Because to be able to do that, means the state is aware of that future too, and could change it. Which would mean it's not definite after all. The excuse "we are not time travelling because I won't tell you the prediction" does not hold. Because if the system allows for the possibility of information to travel, then it becomes a paradox. For me, that would rule out any external mechanism of determinism, even those fundamental to the laws of physics because they would allow for such a possibility.
For example, tell me I'm "defiantly going to the cinema tomorrow" and I will then decide to go to the bowling alley. Now, if you cannot tell me, because it will effect the result, then nothing else can either. If I cannot be told, I cannot measure it. I cannot observe it. So the does the determinism exist?

At least that's how I view the evidence being presented. Sorry if I missed your points, I just see the result as "open" if we get no measurements at all. An open result cannot be reduced to "deterministic" in my view.
It's all physics and stamp collecting.
It's not a particle or a wave. It's just an exchange.

User avatar
tomandlu
Posts: 1111
Joined: Fri Sep 21, 2007 10:22 am UTC
Location: London, UK
Contact:

Re: Wait, can free will exist?

Postby tomandlu » Fri Oct 28, 2011 10:18 am UTC

Technical Ben wrote:For example, tell me I'm "defiantly going to the cinema tomorrow" and I will then decide to go to the bowling alley. Now, if you cannot tell me, because it will effect the result, then nothing else can either. If I cannot be told, I cannot measure it. I cannot observe it. So the does the determinism exist?


This is pretty much the conclusion I came to whilst thinking about this thread yesterday.

It's an interesting (real) paradox. Of course, this only implies that free-will subjectively exists (i.e. an outside observer, in this scenario, could still 'know' what choices the observed subject would make, he just can't tell the subject), but it is arguable that free-will is a purely subjective quality anyway, in that it is an expression of the knowledge possessed by the individual demonstrating free-will, not an expression of all knowledge held in the system. If, for example, the decision to go shopping was going to result in the individual being run-over on the way, then we could hardly argue that we have tested free-will and found that it didn't exist unless we informed the individual and he still decided to go shopping...
How can I think my way out of the problem when the problem is the way I think?

User avatar
doogly
Dr. The Juggernaut of Touching Himself
Posts: 5538
Joined: Mon Oct 23, 2006 2:31 am UTC
Location: Lexington, MA
Contact:

Re: Wait, can free will exist?

Postby doogly » Fri Oct 28, 2011 11:00 am UTC

If Ben can happily admit that Santa, Free Will and Jesus are all equally real, then this is still a great victory for the thread. I'm cool here.
LE4dGOLEM: What's a Doug?
Noc: A larval Doogly. They grow the tail and stinger upon reaching adulthood.

Keep waggling your butt brows Brothers.
Or; Is that your eye butthairs?


Return to “Science”

Who is online

Users browsing this forum: No registered users and 15 guests