excreationist

I think it is possible to reject free will but also believe that criminals should be punished....

If there weren't any punishments for crimes (if they were "forgiven"), people would be less discouraged to commit crimes. So people would have much less reason to avoid speeding, robbing banks, killing people they hate, raping others, etc.

Also, punishing people can alter the person's behaviour so that they don't want to do bad things in the future. (Or at least not get caught)

DRFseven

Yeah, I agree. I was highlighting a "special" (human) reasoning ability typified by this popular analysis. The reasoning that purportedly answers the question is actually answering some other question, because obviously, free will would be true or untrue regardless of whether we "have to" excuse criminals or not. It's kind of like saying, "I don't think we are causing an ozone depletion, because if I did think so, I'd have to wear sunscreen." But, as you know, whatever opinion coincides with the proper neurotransmitter mix is the "correct" opinion to have.

sodium

The problem is that for any anwer there will be a program that gives you this answer. That doesn't mean you'll know to build the right program, but that's your limitation, not a limitation on what is possible for the program.

Any answer is possible, including reasoning and jumping out of the scope of the problem.

Now, you can claim,

1) That's not how our minds work, though. We don't just spit out rote answers.
2) For anything meant to mimic human intelligence, this would fast become impractical.

So, yes, that's true, but what you can't claim is that some problems are non-computational in the sense that no program could solve them, much less that there might be a proof that this is so.

John Page

Just to be clear, I'm not taking this position exactly. The concept of program and computation is open to interpretation as are problem, consciousness etc.

BTW, I'd like to see some form of inverse of the Turing test where the machine has to tell whether its interacting with another machine or human intelligence...

Cheers, john

excreationist

You could train a person who doesn't understand a word of English or perhaps a parrot to do that. i.e. when it hears "duz lyf hav meening?" it parrots back "noh" or "yes" - or a longer answer (exactly as you trained it). That is basically the as same as what you are talking about.
On the other hand, we learn about the world and how to speak over the course of many years and using that experience we generate an answer to that question for ourselves... i.e. we aren't preprogrammed in advance with that question and answer by a programmer or trainer.

sodium

Yes. You see, my argument isn't that a mechanistic process is exactly the same as a human, although I believe that too, and am arguing for it in a different thread. But some people have claimed that they can prove that computers cannot, even in principle, answer certain kinds of questions. Not merely that they can't mimic internal human thought processes, but that they simply can't answer certain questions. I'm pointing out that this is false.

Gillespie

A very interesting thread. Godel has always been a favorite of mine, as relates to the unprovability of knowledge systems. My favorite line of reasoning is the old quote: "Nothing can be created out of nothing." It can. But, define nothing. Seriously think about it. At one point in our past nothing was most likely everything that existed.

God/energy/spirit and nothing. Here in lie the greatest of paradoxes.

I am an atheist. I am most sixty years old. I have experienced a great deal of mechanistic thought. I have also experienced thought far beyond being able to describe as mechanistic. What is woman? How does she work? Why does she work the way she does? How does male/female mechanics/love work? What is karmic control of a life?[ie., schitzophrenia, etc.] What is spiritual hedgmony of entire nations? What is economic suzerainty of all nations by one? How did it happen? Why is it as it is?

If we are ever to solve the AI paradigm, we must resolve the nothing paradox, as well as many others. I have an intuitive feeling AI is most totally paradox dependent. I feel all knowledge is. Nothing exists that can not in some way be paradoxed. [basically, as Godel showed back in the 30's]

I have no proof. I simply believe as Bob Dylan stated: "You want the truth? There ain't no truth." Heisenberg, Einstein, and Godel all contributed greatly to the road we must travel. We have a long way to go to truly and positively know the answers we seek.

mhc

John Page wrote:
THAT, it seems to me, would be the real test that the machine can "do what we do".

sodium

If a computer passes the Turing test, then that means that telling computers from humans is not the kind of thing we can do, so it doesn't make sense to require it of a computer.

John Page

Why not? We use computers to make sense of MRI data, to search for patterns in the myriad of incoming astronomy data.....perhaps we can use computers to tell us who *we* really are.