wu :: forums
        (http://www.ocf.berkeley.edu/~wwu/cgi-bin/yabb/YaBB.cgi)
      

        riddles >> general problem-solving / chatting / whatever >> Revenge of the Right Brain
        
(Message started by: amichail on Jan 27th, 2005, 11:17pm)
Title: Revenge of the Right Brain
Post by amichail on Jan 27th, 2005, 11:17pm
What do you think of this?

http://ideas.web.cse.unsw.edu.au/index.php?module=articles&func=display&ptid=1&aid=433

BTW, I have always tried to put some of the more creative research on CleverCS.
Title: Re: Revenge of the Right Brain
Post by towr on Jan 28th, 2005, 1:27am
Gah..
The extreme polarization between the left/right-brain distinction really irks me. Not all logic/rational thoughts are done by the left brain, nor creativity/emotion by the right. All functions are shared by both halves, and if you loose a hemisphere, though probably impaired, you can still function both rationally and emotionally.
Obviously he's not past 'left-brain' thinking that reduces everything to either/or instead of 'a little bit of both'. Must be one of those people that only uses 10% of his brain, rather than all of it like normal people..
Title: Re: Revenge of the Right Brain
Post by amichail on Jan 28th, 2005, 1:30am
I think the key point of the article is not so much which part of the brain does what, but rather, that creativity will be more valued in the future.
Title: Re: Revenge of the Right Brain
Post by towr on Jan 28th, 2005, 1:43am
Sure, but it really irks me nonetheless.
Of course, at certain times in the past, creativity, artistry, was also much more valued. There's nothing new under the sun..
There will again come a time in the future where rationality is more favored.
Title: Re: Revenge of the Right Brain
Post by amichail on Jan 28th, 2005, 1:47am
Note however that one of the reasons analytic ability will be valued less in the future is due to automation.  Why should we value people who can do proofs when computers can also do proofs?
Title: Re: Revenge of the Right Brain
Post by towr on Jan 28th, 2005, 3:27am
Why should we value people that are creative when computers can also be creative?
It's coming in so far it isn't here yet, and it's already here already up to a point..
Title: Re: Revenge of the Right Brain
Post by amichail on Jan 28th, 2005, 3:35am
If you look at the AI literature, there's a lot more on solving analytic problems than creative ones.

Creativity is also a bit ill-defined, so I suspect that might keep AI researchers away from it for a while.
Title: Re: Revenge of the Right Brain
Post by towr on Jan 28th, 2005, 4:07am
As I said, it already exist up to a point. Ranging from things like chipdesign, to music and even painting.

The point is, there is no great left-right divide. You can't make interesting proofs without some creativity. Even the kind of proofs that computers often have some creative combinations of theorems and rules. You can get there by trial and error, as computers most likely do, or using heuristics, or things like evolutionary computing. But it is in a real sense creative, making new things from old things, in unexpected ways.
Title: Re: Revenge of the Right Brain
Post by amichail on Jan 28th, 2005, 4:17am
I think there is a distinction between creativity for solving well-defined problems (e.g., proving theorems) and creativity required for more open-ended problems (e.g., a painting).

Yes, there has been some research on such things as painting and music composition.  However, I think such research is not as far along as more well-defined analytic problems such as theorem proving say.

In any case, my interest in this topic stems from creativity in more applied computer science research.  In particular, I would like to see new applications unlike anything that we have seen before.

For some reason, much of the existing CS research is very analytic. More creative contributions are often not rewarded as much. Just look at the top conferences/journals.
Title: Re: Revenge of the Right Brain
Post by towr on Jan 28th, 2005, 4:34am
That's because open-ended creativeness in CS is unlikely to result in marketable products. Things like library sort or using binary search on unsorted arrays are fun to play with, but they are not very practical.
Title: Re: Revenge of the Right Brain
Post by amichail on Jan 28th, 2005, 4:44am
Would you say this required creativity?  Is it marketable?

http://www.espgame.org
http://www.captcha.net/esp-search.html
http://www-2.cs.cmu.edu/~biglou/ESP.pdf

What about these?

http://www.idiom.com/~zilla/Work/visualids.pdf
http://www-ui.is.s.u-tokyo.ac.jp/~takeo/java/smoothteddy/index.html
http://berkeley.intel-research.net/paulos/pubs/papers/Familiar%20Stranger%20IR%20Tech%20Report.pdf
http://research.microsoft.com/%7Ehorvitz/cacm-attention.htm

Or hundreds of other ideas on CleverCS?

Open-ended creativeness can be useful I think.
Title: Re: Revenge of the Right Brain
Post by Barukh on Jan 28th, 2005, 4:59am

on 01/28/05 at 01:47:04, amichail wrote:
Why should we value people who can do proofs when computer can also do proofs?

I would say that we value people who can do proofs that computers cannot.

IMHO, computers still can prove only very limited set of (relatively simple) theorems, and it looks like the situation won't change drastically in the future.

Besides, analytic activity may also be very creative (or maybe, ingenious is a better word?)
Title: Re: Revenge of the Right Brain
Post by rmsgrey on Jan 28th, 2005, 7:16am

on 01/28/05 at 01:47:04, amichail wrote:
Note however that one of the reasons analytic ability will be valued less in the future is due to automation.  Why should we value people who can do proofs when computers can also do proofs?

Because there's two aspects to a proof - the manipulation of facts to establish a valid conclusion, and the choice of facts and manipulations to pursue - there are examples on this forum of puzzles whose solution requires a moment of insight in order to start on the right track - for instance, the lower bound for the n-dimensional variant in the Erik's Puzzle thread in the Hard forum is painfuly obvious with hindsight, but I spent months thining about the problem (part-time) and never came close. Someone else spotted that the boundary is non-increasing, and from there the proof's a one-liner.

Yes, a computer could be programmed to take into account any given approach that has worked in the past, but you still need someone to originate those approaches originally - until someone comes up with a computer capable of that sort of intuitive leap, you'll still need humans to work on proofs. Maybe the intuitions that go into proofs can be largely replaced by work-arounds like an exhaustive search of a list of possible approaches, but a lot of the creativity that goes into art can also be replaced - 25-30 years ago, computers were being used to generate abstract images that are qualitatively indistinguishable from some genuine works by an artist whose name escapes me (the difference being that they're works "in the style of" rather than in an original style, but look at the Pop charts...)

Mind you, I'm not convinced that the average man on the street values people who can do proofs anyway...
Title: Re: Revenge of the Right Brain
Post by Sir Col on Jan 28th, 2005, 10:49am
I thought it was an interesting article, and in fairness he did make it clear that the brain worked "in concert", with both sides being responsible for functionality. I felt he was using the notions of "left-sided" and "right-sided" as a metaphor for ideals.

But I'd agree with you, rmsgrey: there is a fundamental difference between applied and pure mathematicians. Because so many higher education courses utlilise mathematics, most people equate these graduates with "mathematicians" or "scientists".

In truth, they are applied scientists and the future of their sciences depends on the brilliance of those strange and rare creatures that are able to work in abstract concepts and then relay this to others in a practical context. Most research scientists/mathematicians produce nothing of lasting signficance. Maybe in the future, computers could be programmed to emulate this type of work, but I doubt if a computer would ever dare to take everything it had been taught and constructively consider the hypothesis, "But if that isn't true..."

Of all the art, poetry, or music ever produced, is there any greater measure of creativity than a real mathematician at work?
Title: Re: Revenge of the Right Brain
Post by JocK on Jan 28th, 2005, 3:44pm

on 01/28/05 at 01:47:04, amichail wrote:
Why should we value people who can do proofs when computers can also do proofs?


Why should we value people who can do proofs when pencils and paper can also do proofs?

Why should we value musicians when a bitstream can also produce music?


Is this a troll or are u serious...?
Title: Re: Revenge of the Right Brain
Post by amichail on Jan 28th, 2005, 3:51pm

on 01/28/05 at 15:44:21, JocK wrote:
Why should we value people who can do proofs when pencils and paper can also do proofs?

Why should we value musicians when a bitstream can also produce music?


Is this a troll or are u serious...?


Actually, pencils and paper can't do proofs.  People use them to do proofs. But computers can do proofs all by themselves.

One may argue that people are better than computers at doing proofs.  I think computers today are probably better than the vast majority of people at proving things.  The vast majority of people probably can't write non-trivial mathematical proofs at all.
Title: Re: Revenge of the Right Brain
Post by Sir Col on Jan 28th, 2005, 4:22pm
I would agree and perhaps go as far to say that the vast majority of mathematicians cannot produce original non-trivial proofs. The crest of most mathematicians' careers is the ability to reproduce accurately complex proofs and, in some cases, possess sufficiently many mathematical experiences to prove a result that they have never seen before. But they will be doing nothing more than reinventing the wheel and making use of someone else's original approach. This, in theory, could be done by a computer; we could "teach" a computer sufficiently many results and allow it to make connections that lead to fruitful outcomes. However, the rare existence of a mathematician who makes that first discovery (which future generations will at best replicate) is beyond A.I.. Simply because it is by definition artificial (simulated), whereas the unique human quality we are talking about is G.I. = genuine intelligence.
Title: Re: Revenge of the Right Brain
Post by Icarus on Jan 28th, 2005, 8:14pm
I'm sorry, but no, no, no, and even no.

I'd thought by now you would of known mathematicians a little better. First of all, you have to produce something original just to become a mathematician in the first place. And you have to keep producing original stuff to stay one (which is why I no longer am a practicing mathematician). This does not mean that the stuff produced is exciting or deep or show "original" thought - just that it has to be something no-one has done before. And just like with the sciences, the vast majority of mathematics produced is uninspired corner-filling. But it is still about doing new stuff, not reproducing old.

Mathematicians are explorers. Mathematics is the undiscovered country they explore. They are trained in areas well-known, but then sent out to see what is beyond the edges. Mostly it turns out to be uninteresting and unhospitable terrain, but occasionally someone finds a pass into a whole new valley filled with wonderful and exciting new places. This is what drives mathematicians - the opportunity to explore new and beautiful ideas. Computers are tools. They have no concept of excitement or beauty. They have nothing to guide them to what is interesting and what is trivial. So even if you could program a computer to prove real theorems (and no-one has, whatever you think), it could never take the place of a mathematician. Not until you give it those so-called "right-brain" skills. If you want a "left-brained" occupation, mathematics is far down the scale.

But I am in full agreement with towr about the original subject. This is complete garbage. It is a false dichotomy of human thought. There is no such thing as a "right-brained" or "left-brained" person. What is perhaps most insulting about this article is the assumption that "foreigners" cannot do "right-brained" activities! Did anyone other than me notice this? The guy says we (the USA) are losing the technical fields to India and other nations who do them just fine, so that other fields they can't do - the so-called "right-brain" fields - are regaining dominance. He whole assumption is that those in India and elsewhere are somehow incapable of doing these jobs as well as we can. Clearly they must of hired an American to design the Taj Mahal! ::)

There are some many places where that essay diverges from reality, it is hard to address them all. The nature of the technical revolution and information age, the causes of outsourcing, and when it is a bad idea, the concept that certain jobs are "left-brained" while others are "right-brained" (I really loved that he included nurses as a "right-brained" job - despite the fact that nurses have to learn more technical information than us supposedly pure-left-brained mathematicians), the concept that people themselves can be divided this way.

People are complex creatures, with a variety of attributes that do not divide properly along these lines. We each come with our own mix of native talents and learned ones. While we tend to gravitate towards jobs that require more of out strong suits, we also tend to modify those jobs as possible (by choosing how we go about them) to more closely align with all our talents. So, given two programmers, you may find one nearly fitting the classic profile of an isolated geek, while the other performs her job with much interaction with others, belying the whole image. To call her "left-brained" is to seriously misjudge her nature. Less obvious, but the same thing is true of the isolated geek. Geeks and nerds in fact have pronounced "right-brain" tendencies that get ignored by those who wish to impose such distinctions. As one who has set on the edge of geek culture, I can tell you that  they have a very well-developed social structure, though one that differs significantly from that of the rest of society.

I could go on for pages, but will let it stand with: that essay is nothing but hot electrons.
Title: Re: Revenge of the Right Brain
Post by amichail on Jan 28th, 2005, 8:22pm

on 01/28/05 at 20:14:24, Icarus wrote:
Computers are tools. They have no concept of excitement or beauty. They have nothing to guide them to what is interesting and what is trivial. So even if you could program a computer to prove real theorems (and no-one has, whatever you think), it could never take the place of a mathematician. Not until you give it those so-called "right-brain" skills. If you want a "left-brained" occupation, mathematics is far down the scale.


In the future, if theorem provers become quite powerful, then mathematicians will indeed focus on guiding those theorem provers to interesting results.  As you say, this is creative.  So indeed, we will value creativity more.
Title: Re: Revenge of the Right Brain
Post by JocK on Jan 29th, 2005, 2:09am

on 01/28/05 at 15:51:57, amichail wrote:
Actually, pencils and paper can't do proofs.  People use them to do proofs. But computers can do proofs all by themselves.


Wow. And I really believed that computers are nothing more than a bunch of switches connected by wires. Silly me.

Can you give an example of a proof done by computers all by themselves?
Title: Re: Revenge of the Right Brain
Post by towr on Jan 29th, 2005, 2:48am

on 01/28/05 at 04:44:22, amichail wrote:
Open-ended creativeness can be useful I think.
But most often only in hindsight. Just like fundamental physics and fundamental mathematics usually does turn out to be very usefull, but only a long way down the line.
There's a tradeoff between shorte term and long term, and businesses and governments too often work on the short term.
Title: Re: Revenge of the Right Brain
Post by towr on Jan 29th, 2005, 2:53am

on 01/28/05 at 16:22:58, Sir Col wrote:
However, the rare existence of a mathematician who makes that first discovery (which future generations will at best replicate) is beyond A.I.. Simply because it is by definition artificial (simulated), whereas the unique human quality we are talking about is G.I. = genuine intelligence.
I disagree, I believe A.I. can, and probably will, one day be G.I.
I don't see any reason why it shouldn't. We're not magical better than other matter.
Title: Re: Revenge of the Right Brain
Post by Sir Col on Jan 29th, 2005, 2:54am

on 01/28/05 at 20:14:24, Icarus wrote:
I'm sorry, but no, no, no, and even no.

Do I detect that you disagree with something I said?  ::)

I believe that, in essence, the mathematical community behaves like a computer. They are trained in the "rules" and "patterns" that have been previously invented/discovered and have been shown to be fruitful. And rightly so!

However, there is not one single example in history where one of the great paradigm shifts in mathematics has occurred without the pioneer facing vehement opposition from the mathematical community at large.

What I was saying is that the pre-requisite for such a move takes great courage and a dogged determination to continue despite the overwhelming criticism, and willingness to "break the rules". I doubt that will ever be achieved by a computer. Who can know for certain?

Do not misunderstand me: I am no expert, as you well know, in A.I., but I suspect that we are a long way away from computers performing the functional/operational tasks of our excellent research mathematicians. Yet I imagine that it can and will be done one day.
Title: Re: Revenge of the Right Brain
Post by towr on Jan 29th, 2005, 3:01am

on 01/28/05 at 20:14:24, Icarus wrote:
So even if you could program a computer to prove real theorems (and no-one has, whatever you think), it could never take the place of a mathematician. Not until you give it those so-called "right-brain" skills.
Oh don't worry, we're working on it ;)
Even so, it still wouldn't replace mathematicians, just create an additional breed.


on 01/29/05 at 02:09:22, JocK wrote:
Wow. And I really believed that computers are nothing more than a bunch of switches connected by wires. Silly me.
And people are 'just' a bunch of atoms. The brain 'just' a bunch of neurons and connections (switches and wires).
Title: Re: Revenge of the Right Brain
Post by amichail on Jan 29th, 2005, 4:17am

on 01/29/05 at 02:09:22, JocK wrote:
Wow. And I really believed that computers are nothing more than a bunch of switches connected by wires. Silly me.

Can you give an example of a proof done by computers all by themselves?


How about this?

http://www.nytimes.com/library/cyber/week/1210math.html
Title: Re: Revenge of the Right Brain
Post by Sir Col on Jan 29th, 2005, 7:28am
Another interesting article; thanks for sharing them, amichail. And don't be put off by the onslaught of criticisms from the members here; they just can't help themselves. In fact, if we hid the names of posters and presented one of their arguments in a similar discussion/thread a few months from now, they'd probably disagree with the points made!  :P

No, seriously, one of the great things about this community is the level of expertise; I know that a good number of the regular posters here rarely play their trump card by admitting their credentials in that particular field. Certainly in most discussions of a scientific nature you can be assured that among all of the members there is at least one person that is currently working, or is in touch with, the cutting-edge ideas from that area/discipline. For example, one of the posters here more than dabbles in the field of A.I..
Title: Re: Revenge of the Right Brain
Post by Icarus on Jan 29th, 2005, 7:43am
Alright, I will assume that the proof is good. Not only computer scientists, but also formalists are estatic about this, for the method used by the computer to prove is formalism. But:

1) The proof the program produced was not a human proof. It took several days effort by a human to convert it into something understandable.
2) If they had put the same effort into producing a program that could create images, I have little doubt that they would have a program by now that could create, with only a little direction for content at the start (much as the choice of theorem to prove is provided to the proving program), original images that invoke the same "artistic" response that human paintings do. In fact, this is almost possible now.
3) As has already been said, even when fully developed, this and similar programs will not at all replace the need for mathematicians. Rather, it will simply provide another tool in the mathematician's toolbox. Contrary to what the article suggests, it will not put an end to mathematicians proving things themselves either. In order for mathematicians to guide development of a theory, they need a feel for what is possible and what is not. The only way to get this feel is to understand how proofs in the theory work, how things inter-relate. The only way to get this feel is with hands-on experience. By the time programs reach the state where they can make the sorts of decisions required to develop a theory themselves, they will have achieved true intelligence.



Sir Col - only one of the "no"s was for you and your pronouncement that most mathematicians simply re-iterate proofs rather than producing new ones. The rest were for the whole false dichotomy between "left-brain" and "right-brain". While it is true that most mathematicians do not produce major new theories, this is to be expected. What leads to major new discoveries is only partly intellect and creativity. A large portion is simply the luck of having chosen the right problem or problems to work on, to suggest the course leading to the discovery.

And I think you greatly overstate the problem with the "establishment" repressing the new idea. Far from repression, many new ideas are granted quick, wide-spread acceptance. Consider the two greatest "paradigm-shifts" (a terminology I find almost as inappropriate as "left-brained" and "right-brained") in the history of mathematics: The introduction of the idea of "proof" by Thales of Miletus, and the introduction of analytic geometry by Rene DeCartes. I have not heard of Thales having to overcome opposition to his idea. Indeed, as a merchant, he probably would not have bothered with it if his ideas had met great opposition. His mathematics was only a hobby. And analytic geometry was met with immediate acceptance, and even great joy by mathematicians of DeCartes day.

I have not heard of any great repression of Calculus when it was introduced, either. There was a lot of hostility, but it was over who got to claim credit. There was also resistance, but this resistance was based on very sound mathematical reasoning. Calculus as introduced was full of fuzzy thinking and ludicrous concepts (much of which are still to be found in the 0.999... thread;)). It would take over a century before all the problems were ironed out, and Cauchy set calculus on a firm footing. Resistance, though, is not the same thing as repression.

While mathematical history has its share of bitterness and repression, it is far from universal.
Title: Re: Revenge of the Right Brain
Post by THUDandBLUNDER on Jan 29th, 2005, 7:45am
Yeah, thank you for sharing, amichail.  :'(

Icarus, Sir Col is probably referring to the persecution, both real and later merely perceived, of Cantor. And by 'resistance' to the concepts of calculus, I take it that you mean that of Bishop Berkeley, whose perceptions of 'truth' were probably influenced as much by his strong religious beliefs and instilled awe of the infinite as by any innate hunger for mathematical objectivity. As for your abhorrence of 'paradigm shifts', I too believe that they belong more to the subjective realm of the senses than to the objective world of pure ideas.


on 01/29/05 at 03:01:01, towr wrote:
And people are 'just' a bunch of atoms. The brain 'just' a bunch of neurons and connections (switches and wires).

And in what year do you think computers will become aware (of their own existence) in the sense that humans are? In other words, when will the Halting Problem be replaced in university textbooks by the Terminating Problem? (LOL) Do you believe that there is a gene for consciousness?

Title: Re: Revenge of the Right Brain
Post by Icarus on Jan 29th, 2005, 7:57am

on 01/29/05 at 07:28:30, Sir Col wrote:
And don't be put off by the onslaught of criticisms from the members here; they just can't help themselves.


I had not thought of my posts as being critical of amichail. He asked for our thoughts, and we have been giving them. By replying to his posts, and each others, and by his replying to ours, we each sharpen our own concepts and clarify our thinking on the matter. The freedom to say "this is wrong", and explain why we believe so, without having to worry about hurt feelings is crucial to an open discussion and the search for truth. (Insults are another matter - if someone cannot defend their own position without insulting their opposition, then they have no position worth defending. But no one in this thread has been insulting that I have noticed.)
Title: Re: Revenge of the Right Brain
Post by JocK on Jan 29th, 2005, 9:49am

on 01/29/05 at 03:01:01, towr wrote:
And people are 'just' a bunch of atoms. The brain 'just' a bunch of neurons and connections (switches and wires).

Your first statement is plainly wrong. The second I agree with.
Title: Re: Revenge of the Right Brain
Post by Sir Col on Jan 29th, 2005, 9:54am
T&B, I wasn't just thinking about Cantor's treatment, but I believe that it occurs at the genesis of all the greatest "transistions". For example, we only read about the praise and accolades bestowed on the inventors of the non-Euclidean geomteries. However, prior to this birth we know from some of Gauss's letters that the he was fully aware that no contradiction would exist by adapting the fifth postulate. In fact, he secretly developed a consistent geometry where the sum of angles in a triangle are less than two right angles. But through fear of how it may have damaged his reputation he was reluctant to publish. After all, no credible mathematician would question the soundness of Euclidean geometry!

Icarus, my only reason for saying that was to warn amichail that it was not personal. Yet to someone new, every single post you make being opposed by one or two other people could be interpreted as such; as regulars, we are fairly immune to the challenges as we know the heart (and expertise) behind each retort. We have already discussed this as being one of the fine qualities of this community, but sometimes we may need to be a little less aggressive with new members. But once we've lured them in, then we can go for the kill! ;D
Title: Re: Revenge of the Right Brain
Post by JocK on Jan 29th, 2005, 10:19am

on 01/29/05 at 04:17:00, amichail wrote:
How about this?

http://www.nytimes.com/library/cyber/week/1210math.html


Yup... automated deductive reasoning. A nice example of the effective use of computers as a tool helpful to mathematicians. Just like pencil & paper.

Amichael, I think you are using the wrong examples to prove something that is obvious to all of us. Of course it is only creativism that counts. Nothing new here, this is known for centuries. What scientific breakthroughs relied solely on deductive reasoning? Newton's laws, Maxwell's electromagnetism, Einstein's general relativity, Dirac's relativistic wave equation, Feynman's path integral interpretation of quantum mechanics: all perfect examples of mindblowingly beautiful inductive reasoning. True creativity at work.

Sorry my friend, but computers are (and will remain) in an slightly different league. (Seems to me the biggest understatement I made in years... ;D )
Title: Re: Revenge of the Right Brain
Post by amichail on Jan 29th, 2005, 1:59pm

on 01/29/05 at 10:19:41, JocK wrote:
Yup... automated deductive reasoning. A nice example of the effective use of computers as a tool helpful to mathematicians. Just like pencil & paper.

Amichael, I think you are using the wrong examples to prove something that is obvious to all of us. Of course it is only creativism that counts. Nothing new here, this is known for centuries. What scientific breakthroughs relied solely on deductive reasoning? Newton's laws, Maxwell's electromagnetism, Einstein's general relativity, Dirac's relativistic wave equation, Feynman's path integral interpretation of quantum mechanics: all perfect examples of mindblowingly beautiful inductive reasoning. True creativity at work.

Sorry my friend, but computers are (and will remain) in an slightly different league. (Seems to me the biggest understatement I made in years... ;D )


I don't think there is anything obvious about this discussion.  

First, it's not obvious how a mathematician thinks.  How does creative and analytic thought mix? Mathematics is not forgiving, yet somehow mathematicians have a "feeling" for what is true and worth pursuing, the general direction that a proof could take, etc.  There is nothing obvous in this.

Second, what happens to the mathematician's thought patterns in the presence of a very powerful theorem prover?  Would they change much?  If they can become more creative, does this mean someone with weaker analytic ability but strong creativity could do interesting mathematics?

Third, an analogy with chess might be helpful.  Is chess really so interesting now given that computers can pretty well beat any human player?  Or is the game more interesting when the computer is used as a tool to help people play more effectively? How would the human's thought patterns change when playing chess using a powerful computer helper?

Fourth, given these very powerful thought tools, should we teach people differently?  Should we focus more on vague creative concepts?
Title: Re: Revenge of the Right Brain
Post by amichail on Jan 29th, 2005, 2:05pm

on 01/29/05 at 07:43:32, Icarus wrote:
Alright, I will assume that the proof is good. Not only computer scientists, but also formalists are estatic about this, for the method used by the computer to prove is formalism. But:

1) The proof the program produced was not a human proof. It took several days effort by a human to convert it into something understandable.


Why does it have to be understandable?  Isn't it sufficient that the program that produced it is proven correct so that we know that the theorem is true?

The following might be of interest:

http://ideas.web.cse.unsw.edu.au/index.php?module=articles&func=display&ptid=1&catid=13&aid=151

BTW, why is it that people think theorems are true for interesting reasons?  What if the reasons are not so interesting?  Would we still be interested in understanding the proof?  What if no insightful proof exists?
Title: Re: Revenge of the Right Brain
Post by rmsgrey on Jan 29th, 2005, 2:31pm

on 01/29/05 at 14:05:34, amichail wrote:
Why does it have to be understandable?  Isn't it sufficient that the program that produced it is proven correct so that we know that the theorem is true?

This is one of my biggest problems with the classic form of the Turing Test - while a computer that can convince me that it's human will certainly convince me that it's intelligent, it seems very arrogant to assume that intelligent beings must think exactly the same way humans do...

And even some human mathematicians write their proofs in ways that require translation to be understandable by modern mathematicians - if you look at "textbook" proofs from a couple of centuries ago, it's not unusual for them to be almost incomprehensible.

Just because something isn't phrased the same way as I would phrase it doesn't mean that it's not of the same order of value as anything I could produce.
Title: Re: Revenge of the Right Brain
Post by Icarus on Jan 29th, 2005, 8:50pm

on 01/29/05 at 14:05:34, amichail wrote:
Why does it have to be understandable?  Isn't it sufficient that the program that produced it is proven correct so that we know that the theorem is true?


First, my point was that the computer is still not accomplishing the same thing as the human mathematician. I myself can program a computer to make all sorts of pictures based solely on  its own "volition". I could probably program it to produce pictures more appealing than some so-called "modern art". But I cannot program it to make pictures that anyone would care to see, much less own. In the same way, what the computer is producing is far different than what the mathematician does.

Second, whether the computer proof is enough depends on what you want. Most people believe mathematics is nothing more than memorizing formulas by rote, and applying them mechanically. This is essentially what the program does. It has a set of rote strategies, and applies them mechanically until it finds a proof. But those formulas that students learn and use by rote speak volumes to the mathematician about the nature of the theory they are part of. The student uses them, but the mathematician understands them.

There are so many similes that come to my mind for this that it is hard to choose one. But this one is closest to my own soul: Regularly I hear the argument that we should abandon manned space flights as too expensive and dangerous. We are told that every thing we could learn from manned exploration we can also learn from unmanned probes, but for much less money. I hold that this is a terribly short-sighted argument. What is the purpose of all this information about other places if we never intend to go there. Some may be satisfied with pictures and data. As for me, I want to walk on Mars.

So also others may be satisfied simply to know that every Robbins Algebra is Boolean. As for me, I want to know why.


Quote:
The following might be of interest:


Not really (but thanks for the link, and the discussion). There is nothing there that is new to me. It (almost) totally misses my point. I do not hold that a computer-assisted proof like the 4-color theorem or computer-generated proof like the Robbins Algebra theorem are any less valid than a normal proof. I have said as much about the 4-color theorem before. I merely suggest that they are less useful, less informative. They are the pictures and data sent back by probes. Interesting, useful to build upon, but in the end, not what I desire.


Quote:
BTW, why is it that people think theorems are true for interesting reasons?  What if the reasons are not so interesting?  Would we still be interested in understanding the proof?  What if no insightful proof exists?


Then what is the point of studying it? A good proof cannot help but offer insight. The Robbins proof had the insight - it was just difficult to see in the original form. It took several days of distilling by a human to bring the insight to the fore. The 4-color theorem took a great deal of insight to get to the point where the computer could help out at all. And even then, "because each case worked" offers its own insights into the nature of the problem.

We are interested in these insights because they tell us what to look at next. Another simile: In my group at work, everyone knows how to use computers to do our jobs. But some, such as myself, know how computers work, and how to make them work for us. Those who do not understand are limited in what they can do to the tools that have already been prepared for them. I, on the other hand, have developed a reputation for doing things no one else can, because I know how to program my computers to do it. (We use both PC and UNIX machines, and no one else in the group has near the UNIX skills that I have - though the IT guy who assists our group leaves me in his dust.)

If we decide not to bother with the insights behind the theorems, we become like the non-programming members of my workgroup. They are solid workers, who are good at their jobs, but they cannot move beyond their current limits without someone else to show them the way. I prefer to lead the way. But to do that, I need to understand what I am working with. To lead, I need the insights to be found in the proofs.



on 01/29/05 at 14:31:57, rmsgrey wrote:
This is one of my biggest problems with the classic form of the Turing Test - while a computer that can convince me that it's human will certainly convince me that it's intelligent, it seems very arrogant to assume that intelligent beings must think exactly the same way humans do...


rmsgrey - You misunderstand the Turing Test. It was offered as a sufficient condition, not a necessary one.

The whole point of the Turing test was that if a machine passed it, you could not deny its sentience without denying that of other people, and ultimately of yourself. Your comments suggest that it also stated that machines failing the test were not to be considered sentient. This is not the case. A machine failing the test may or may not be sentient. When the test fails, it is the test that failed, not the machine. There was no assumption that all intelligences must think like us.
Title: Re: Revenge of the Right Brain
Post by JocK on Jan 30th, 2005, 4:10am

on 01/29/05 at 13:59:37, amichail wrote:
... what happens to the mathematician's thought patterns in the presence of a very powerful theorem prover?  Would they change much?  If they can become more creative, does this mean someone with weaker analytic ability but strong creativity could do interesting mathematics?


What happened to the mathematician's thought pattern following the invention of pen & paper?
-- Nothing at all, most mathematicians became more efficient though.

What happened to the mathematician's thought pattern following the invention of symbolic manipulation programs?
-- Nothing at all, many mathematician started working more efficiently though.

What will happen to the mathematician's thought patterns when very powerful theorem provers become available?
-- Nothing at all, but surely many mathematicians will start working more efficiently.


Again, it's all about inductive reasoning. Brainwaves based on insight, and (yes!) creativism. All the rest doesn't count and can (should!) be automated.

Title: Re: Revenge of the Right Brain
Post by rmsgrey on Jan 30th, 2005, 6:49am

on 01/29/05 at 20:50:43, Icarus wrote:
rmsgrey - You misunderstand the Turing Test. It was offered as a sufficient condition, not a necessary one.

The whole point of the Turing test was that if a machine passed it, you could not deny its sentience without denying that of other people, and ultimately of yourself. Your comments suggest that it also stated that machines failing the test were not to be considered sentient. This is not the case. A machine failing the test may or may not be sentient. When the test fails, it is the test that failed, not the machine. There was no assumption that all intelligences must think like us.

The original argument by Turing is one thing, but the way the average layman refers to the Turing Test is as a necessary condition for machine intelligence, rather than a sufficient one. My use of the word "classic" to describe this common meaning of the phrase may have been inappropriate, but I was seeking to clarify the distinction between the traditional Turing Test and the various variations that have been proposed that test for whether or not a machine can convince us of its intelligence rather than its humanity.
Title: Re: Revenge of the Right Brain
Post by towr on Jan 30th, 2005, 7:46am

on 01/29/05 at 07:45:49, THUDandBLUNDER wrote:
And in what year do you think computers will become aware (of their own existence) in the sense that humans are? In other words, when will the Halting Problem be replaced in university textbooks by the Terminating Problem? (LOL)
I have no idea when, and if, that will be. I just don't see a reason why it can't be. There is no magic, and if there is, no reason why it can't effect computers.
If god (if you're inclined to believe in him) wants to grant computers souls (if you're inclined to believe in them), that's his business.


Quote:
Do you believe that there is a gene for consciousness?
Not one, at least. Certainly many genes that are prerequisite (in humans) for the devellopment of one. We need a brain to get a consciousness, and genes to get a brain. If that is all that there is to it, brain=mind, then so be it.
Title: Re: Revenge of the Right Brain
Post by towr on Jan 30th, 2005, 7:48am

on 01/29/05 at 09:49:32, JocK wrote:
Your first statement is plainly wrong. The second I agree with.
How so is it wrong? Are we composed of more than atoms?
Well, granted, we're atoms, and how they are 'connected'. (Just like a computer is not just its parts, but also how it's put together) But aside from it?
Title: Re: Revenge of the Right Brain
Post by towr on Jan 30th, 2005, 8:06am

on 01/29/05 at 20:50:43, Icarus wrote:
The whole point of the Turing test was that if a machine passed it, you could not deny its sentience without denying that of other people, and ultimately of yourself.
I don't think I ever heard it put so well.
Of course, people being people, deny it anyway. Even though nothing non-human has even ever passed the turing test, they're already denying it solely on the grounds that "it's a computer".
Frankly it doesn't matter if we devellop a computer that is creative, left-brained, right-brained, prodigious, genius, original in all it's ideas and expressions. To many people, a priori, it's not human, so it's not intelligent. It misses the magic.
Never mind you could say the same thing about other people. How do we know they're not just automatons?
Title: Re: Revenge of the Right Brain
Post by JocK on Jan 30th, 2005, 10:36am

on 01/30/05 at 07:48:47, towr wrote:
How so is it wrong? Are we composed of more than atoms?
Well, granted, we're atoms, and how they are 'connected'. (Just like a computer is not just its parts, but also how it's put together) But aside from it?


Would you mind doing the following 'experiment' then? Lay down on the floor... put a plastic bag tightly over your head, and... wait a few minutes.*

All your atoms are still there. No connections were removed. Still the same you.

Or is it?

Despite all scientific advances, we should not pretend we even have the slightest understanding of what actually is this magical 'thing' that makes us different from 'just' bunches of switches. We have given it a name - consciousness - and we know some bunches of interconnected switches (e.g. humans) do have it, and other bunches of switches (e.g. computers) will never get it. Apart from that.... we lack any basic understanding on the subject.


* Please, don't be offended by this gedankenexperiment. And... don't try this at home: in contrast to that bunch of switches that you have powered on a few minutes ago, you might not be reboot-able!!  ;)
Title: Re: Revenge of the Right Brain
Post by towr on Jan 30th, 2005, 11:03am

on 01/30/05 at 10:36:50, JocK wrote:
Would you mind doing the following 'experiment' then? Lay down on the floor... put a plastic bag tightly over your head, and... wait a few minutes.*
And? If you pull the plug of the computer from its socket it also 'dies'. Admittedly you can restart that one more easily. But it's a case of losing power; no oxygen means no energy to function. And in the case of a body that causes damage that isn't reversible unless you act quickly.


Quote:
All your atoms are still there. No connections were removed. Still the same you.

Or is it?
Obviously not. All sorts of chemical reactions are halted. Homeostasis, part of the functional configuration, is broken. Cells start dying (definatly involving bonds between atoms being destroyed) etc


Quote:
Despite all scientific advances, we should not pretend we even have the slightest understanding of what actually is this magical 'thing' that makes us different from 'just' bunches of switches.
So we should just pretend it is really magic? And not natural/physical?


Quote:
We have given it a name - consciousness - and we know some bunches of interconnected switches (e.g. humans) do have it, and other bunches of switches (e.g. computers) will never get it.
How do you know that?
What more is it than a guess or a wish, a desire to be special?
Title: Re: Revenge of the Right Brain
Post by rmsgrey on Jan 30th, 2005, 11:11am

on 01/30/05 at 10:36:50, JocK wrote:
Despite all scientific advances, we should not pretend we even have the slightest understanding of what actually is this magical 'thing' that makes us different from 'just' bunches of switches. We have given it a name - consciousness - and we know some bunches of interconnected switches (e.g. humans) do have it, and other bunches of switches (e.g. computers) will never get it. Apart from that.... we lack any basic understanding on the subject.

Saying that a given bunch of switches can never become conscious requires a lot of understanding of what consciousness is and requires. If you don't know what is necessary for consciousness to develop, how can you ever be sure that the biro in my pocket could never become conscious, let alone a device that is capable of mimicking thought?

If you can say what quality it is that humans have that allows them to become conscious, and why lacking it prevents computers from becoming conscious, then you understand consciousness a lot better than most people will admit to.
Title: Re: Revenge of the Right Brain
Post by JocK on Jan 30th, 2005, 11:20am

on 01/30/05 at 11:11:25, rmsgrey wrote:
how can you ever be sure that the biro in my pocket could never become conscious <..> ?


Yeah... let's start the Biro Liberation Front..!! We have deprived these conscious beings from their 'human' rights way too long..!  ;)
Title: Re: Revenge of the Right Brain
Post by JocK on Jan 30th, 2005, 11:59am

on 01/30/05 at 11:03:41, towr wrote:
And? If you pull the plug of the computer from its socket it also 'dies'. Admittedly you can restart that one more easily. But it's a case of losing power; no oxygen means no energy to function. And in the case of a body that causes damage that isn't reversible unless you act quickly.



OK, let's use another - perhaps more illuminating - gedankenexperiment.

Using a newly developed hi-res body scanner/synthesizer I am able to register for any device the positions of all atoms and each electron, and with that information I can construct an exact copy of the device.

Now, I first scan your computer, and subsequently build an exact copy. I put the power on and it works. More amazingly, the memory content is the same, and all your files are there. Next, I destroy your original computer, and give you the copy I just created. If I don't tell you what happened, would you ever notice? I don't think so.

Now I scan another device.. YOU..! The result of the scanning and synthesizing is a second Towr. Exactly the same atoms, connected in the very same way. I then approach you and tell you... "sorry, have to switch you off...". Would you mind? Would your 'whole being' be represented in the copy-Towr?


on 01/30/05 at 11:03:41, towr wrote:
So we should just pretend it is really magic? And not natural/physical?


Absolutely.

I agree there is a slight chance that in some distant future profound new insights in quantum physics might ever lead to a level of understanding of consciousness in terms of 'physical laws'. At this moment in time however, we have not even scratched the surface of the subject. And yes, as long as we don't have the faintest understanding of the 'trick', it is magic to all of us.
Title: Re: Revenge of the Right Brain
Post by rmsgrey on Jan 30th, 2005, 12:35pm

on 01/30/05 at 11:59:47, JocK wrote:
OK, let's use another - perhaps more illuminating - gedankenexperiment.

Using a newly developed hi-res body scanner/synthesizer I am able to register for any device the positions of all atoms and each electron, and with that information I can construct an exact copy of the device.

Now, I first scan your computer, and subsequently build an exact copy. I put the power on and it works. More amazingly, the memory content is the same, and all your files are there. Next, I destroy your original computer, and give you the copy I just created. If I don't tell you what happened, would you ever notice? I don't think so.

Now I scan another device.. YOU..! The result of the scanning and synthesizing is a second Towr. Exactly the same atoms, connected in the very same way. I then approach you and tell you... "sorry, have to switch you off...". Would you mind? Would your 'whole being' be represented in the copy-Towr?

But if you copy Towr perfectly, and then destroy the original, I wouldn't ever notice that if you didn't tell me. If you duplicated yourself perfectly, and then destroyed the original, but didn't tell Towr, would Towr ever notice? Would he care if you did tell him?

What your thought experiment establishes so clearly is that there is a difference between "me" and anything else. In the case where I am duplicated, there is a significant difference between me and my copy - one is me; the other is him. I object strenuously to terminating "me" (except under certain conditions) but have no more qualms about terminating "him" than about terminating "JocK".

If you were to render me unconscious, duplicate me perfectly, destroy the original, and then wake me up, I might be a little upset about the rendering unconscious part, but wouldn't much care about the duplication and destruction...
Title: Re: Revenge of the Right Brain
Post by JocK on Jan 30th, 2005, 12:50pm

on 01/30/05 at 12:35:25, rmsgrey wrote:
But if you copy Towr perfectly, and then destroy the original, I wouldn't ever notice that if you didn't tell me.


Are you sure? Would the copy of Towr have Towr's consciousness?

Suppose I scan Towr, put all information on a memory stick, and destroy Towr before constructing the copy. Where is Towr's consciousness. In my memory stick?
Title: Re: Revenge of the Right Brain
Post by towr on Jan 30th, 2005, 12:52pm

on 01/30/05 at 11:59:47, JocK wrote:
Now I scan another device.. YOU..! The result of the scanning and synthesizing is a second Towr. Exactly the same atoms, connected in the very same way. I then approach you and tell you... "sorry, have to switch you off...". Would you mind? Would your 'whole being' be represented in the copy-Towr?
Yes, but also in 'me'-me. And 'me'-me doesn't want to die anymore than 'copy'-me.
Of course if you just killed 'me'-me, 'copy'-me wouldn't know any better, no more than I could tell the difference between my original computer and an exact copy.
But then, I wouldn't want to lose many things I don't have and only believe I have either. The 'me'-ness can be entirely ficticious, and I'd still mind losing it. Really, what should I care if I'm dead, certainly I won't mind anymore then. Either I'm not, or somewhere better (hopefully).


Quote:
I agree there is a slight chance that in some distant future profound new insights in quantum physics might ever lead to a level of understanding of consciousness in terms of 'physical laws'. At this moment in time however, we have not even scratched the surface of the subject. And yes, as long as we don't have the faintest understanding of the 'trick', it is magic to all of us.
It seems like magic. Just as anything people don't understand. TV is 'magic' in that way to some. But it's still just physics, unknown perhaps, or misunderstood, but still, just physics. Following set rules of nature.
I'm pretty sure most computer programs are magic to a lot of people. So what's the problem anyway? As long as it's too complex to understand it can be sentient, right? So just a matter of reaching the right amount of complexity.

And even aside from that. Since when did we ever need to understand something to make it? We've been making babies since before we could conceive the concept of conception. And fire before we knew what oxidation was. Bread before we knew yeast. etc Artificial evolution may well lead to things we can never design intentionally. Artificial life not being the least of them.
Magic!
Title: Re: Revenge of the Right Brain
Post by towr on Jan 30th, 2005, 12:55pm

on 01/30/05 at 12:50:53, JocK wrote:
Are you sure? Would the copy of Towr have Towr's consciousness?
Yes


Quote:
Suppose I scan Towr, put all information on a memory stick, and destroy Towr before constructing the copy. Where is Towr's consciousness. In my memory stick?
Frankly, given the chance I might do it myself, just as a backup you know. And in the event of my death have my loyal minions create a new android body and upload me back to life.
As every CS minded person knows, you should make frequent backups of everything important ;)

[e]To be more precise, you'd have a physical representation of my consiousness at a point in time. Consciousness, like most mental 'states' are processes, so my representation of consciousness on the memorystick isn't conscious.[/e]
Title: Re: Revenge of the Right Brain
Post by rmsgrey on Jan 30th, 2005, 1:30pm

on 01/30/05 at 12:50:53, JocK wrote:
Are you sure? Would the copy of Towr have Towr's consciousness?

If it didn't, then it wouldn't be a perfect copy, now would it?

Quote:
Suppose I scan Towr, put all information on a memory stick, and destroy Towr before constructing the copy. Where is Towr's consciousness. In my memory stick?

If I write a program in Java, then it's compiled to run on a virtual machine. While the PC I'm running the Java program on isn't running the JVM, where is the virtual machine? On the hard disk?

Or while I'm not playing Starcraft, where are all the Zerglings?

Or how come I can't live in a blueprint of my house?

If Towr's consciousness is a property of the interactions of his component particles (under the laws of phsyics) then his consciousness only exists while they're interacting. If you were to somehow simulate Towr for a while using the copied information, then you'd have an interesting question about where his consciousness was
Title: Re: Revenge of the Right Brain
Post by amichail on Jan 30th, 2005, 2:20pm

on 01/30/05 at 12:50:53, JocK wrote:
Are you sure? Would the copy of Towr have Towr's consciousness?

Suppose I scan Towr, put all information on a memory stick, and destroy Towr before constructing the copy. Where is Towr's consciousness. In my memory stick?


Let's suppose that people have a consciousness -- that they are more than just a bunch of atoms.

How does this show that people should play a key role in constructing proofs rather than leaving them for computers?

Perhaps you might appeal to undecidability?  

But even then, you need to consider the average performance of people vs computers.  
Title: Re: Revenge of the Right Brain
Post by amichail on Jan 30th, 2005, 2:26pm

on 01/29/05 at 20:50:43, Icarus wrote:
There are so many similes that come to my mind for this that it is hard to choose one. But this one is closest to my own soul: Regularly I hear the argument that we should abandon manned space flights as too expensive and dangerous. We are told that every thing we could learn from manned exploration we can also learn from unmanned probes, but for much less money. I hold that this is a terribly short-sighted argument. What is the purpose of all this information about other places if we never intend to go there. Some may be satisfied with pictures and data. As for me, I want to walk on Mars.

So also others may be satisfied simply to know that every Robbins Algebra is Boolean. As for me, I want to know why.


I like this analogy with manned space exploration.

Manned space exploration seems to appeal more to emotion than scientific necessity.

One reason for having mathematicians stay in the loop -- rather than deligate proofs to computers -- is because it is more rewarding for them.

Note this however: I can imagine a world where the majority of proofs are done by computers, yet mathematicians still stay in the loop in a small number of proofs to gain insight.  In this way, mathematics would perhaps progress more rapidly as a search for truth, yet the field would remain rewarding for mathematicians.
 
Title: Re: Revenge of the Right Brain
Post by Grimbal on Jan 30th, 2005, 4:41pm
Why should musicians still play Beethoven's 9th symphony since we already have numerous records of it?
Title: Re: Revenge of the Right Brain
Post by rmsgrey on Jan 31st, 2005, 8:52am

on 01/30/05 at 16:41:40, Grimbal wrote:
Why should musicians still play Beethoven's 9th symphony since we already have numerous records of it?

1) There's a difference between doing and witnessing. Musicians playing the piece get something different from the experience than anyone just listening to it.

2) Unlike a recording, where (apart from gradual degradation) each performance is objectively identical, each live performance is subtly different, and has the potential for something completely new.

3) People are prepared to pay them to do it. And that's partly due to live performances being more expensive than recordings, so having more snob value...
Title: Re: Revenge of the Right Brain
Post by JocK on Jan 31st, 2005, 3:33pm


on 01/30/05 at 13:30:51, rmsgrey wrote:

If it didn't, then it wouldn't be a perfect copy, now would it?


Well, it would be a perfect copy without the 'magic'...



on 01/30/05 at 13:30:51, rmsgrey wrote:
If Towr's consciousness is a property of the interactions of his component particles (under the laws of phsyics) then his consciousness only exists while they're interacting. If you were to somehow simulate Towr for a while using the copied information, then you'd have an interesting question about where his consciousness was


You start with a very big IF I think: perhaps you can simulate the dynamics of all atoms constituting Towr, but how would you simulate the 'magic' associated with that 'bunch of switches'? We simply have no clue how consciousness gets associated with certain 'bunches of switches'.

Yet, despite our absolute lack of understanding, we do have a whealth of empirical evidence that tells us that manmade 'bunches of switches' never manifest any signs of consciousness. And yes, I know there are folk (some of them who perhaps have seen too many Frankenstein movies?) who tell themselves that it's just a matter of scale ("If we keep adding switches, at some stage it will become alive and conscious!").

Reminds me a bit of those who waist their lives trying to construct perpetual motion machines violating no fundamental microscopical physical laws other than that somewhat strange statistical observation called the second law of thermodynamics.

Title: Re: Revenge of the Right Brain
Post by Icarus on Jan 31st, 2005, 5:56pm
Calling it "consciousness" does not evade or any way change the religious aspect of it. What you are talking about is the "soul". The question is: do individuals have a "soul" with an existance beyond the material properties of the body. If not, then the copy will be a second towr, diverging from the original as he obtains different experiences. If so, then a second question arises as to whether the soul is copied, or a new soul is created, or the duplicate towr is souless.

The fact is, these questions cannot be answered scientifically at this time. You can extrapolate from testable results, but the extrapolation is still to far to be reliable. Until non-human sentience is available, or the ability to copy humans as described, experiments cannot be performs that would shed light on the question.

Therefore all the statements made in this regard should be understood as being opinion, and not based on hard evidence. I.e., you are not likely to convince each other of your point of view, and would be wise to simply agree to disagree.


on 01/31/05 at 15:33:54, JocK wrote:
Yet, despite our absolute lack of understanding, we do have a whealth of empirical evidence that tells us that manmade 'bunches of switches' never manifest any signs of consciousness. And yes, I know there are folk (some of them who perhaps have seen too many Frankenstein movies?) who tell themselves that it's just a matter of scale ("If we keep adding switches, at some stage it will become alive and conscious!").


Some of the older generation of science fiction writers (R. Heinlein in particular) suggested something like this for a while (see, for example, "The Moon is a Harsh Mistress"). But this was because they were writing about something they had no conception of. Anyone holding a similar view today also has no real conception of how computers work, and probably much less an idea of how the brain works than is currently available (great inroads into this problem have been made in the last 30 years or so). Likely, they have also been influenced by these old S.F. stories, without learning how outdated and ludicrous they were.

True A.I., if and when it comes, will come from solid programming, not from simply tossing more switches together. To expect that would be similar to expecting a cathedral to appear once you threw enough boards and rocks into a pile. And claiming that because no computer yet has shown sentience, it is strong evidence that A.I. is impossible, is also akin to claiming cathedrals are impossible because piles of rock and board do not turn into them all on their own.

I do not claim that true A.I. is possible. I also do not claim it is impossible. The evidence is just not there yet, either way.
Title: Re: Revenge of the Right Brain
Post by towr on Feb 1st, 2005, 12:59am

on 01/31/05 at 17:56:53, Icarus wrote:
True A.I., if and when it comes, will come from solid programming, not from simply tossing more switches together.
Certainly not the latter, but what to you mean by the former?
F.i. in how far are neural networks and evolutionary computing, 'solid programming'? Because in those cases you're obviously not programming the eventual overt behaviour.
Title: Re: Revenge of the Right Brain
Post by rmsgrey on Feb 1st, 2005, 8:23am
[quote author=Icarus link=board=riddles_general;num=1106896655;start=50#56 date=01/31/05 at 17:56:53]Some of the older generation of science fiction writers (R. Heinlein in particular) suggested som ÿÿÿðE-ñw7Ç`
Title: <ÿÿàÿÿü
Post by on
Title: €ÿÿþ @ 7È 7Ç`€ÿÿÿÿÿÿÿÿÿÿÿÿ6ø«Áÿÿ\`Ty of getting enough rock and board together you can't build a cathedral. Or if you can't get enough Uranium together, you can't build an atomic bomb.

There are some stories (I think Arthur C Clarke wrote a short about the global phone exchange waking up) where a merely complex system spontaneously wakes up, but I'm drawing a blank trying to think of any other examples of machines that spontaneously wake up rather than being designed as AI.

[quote]Calling it "consciousness" does not evade or any way change the religious aspect of it. What you are talking about is the "soul". The question is: do individuals have a "soul" with an existance beyond the material properties of the body. If not, then the copy will be a second towr, diverging from the original as he obtains different experiences. If so, then a second question arises as to whether the soul is copied, or a new soul is created, or the duplicate towr is souless.[/quote]
I don't know what a soul is, or even if I have one. I talk about consciousness because I know I have it - though I can't know whether anyone else actually is conscious. If you define a soul as being one's consciusness, then, yes, that's what I've been discussing. If you define a soul as "the immortal part of a mortal being" then I have no idea whether such a thing really exists. If you define it as "African American ethnic pride" then it's definitely not what I've been considering.

Yes, discussions about personal identity, consciousness and so on can move into religious issues. That's what I was trying to dodge by specifying a "perfect" copy. Since the mechanism of copying as described was already invoking some form of "magic" to get away with an apparent impossibility (knowing the complete state of every particle and duplicating it accurately) including a hypotheticl soul by an equally undisclosed mechanism seemed not too much of a stretch.
Post by on
Title: Re: Revenge of the Right Brain
Post by THUDandBLUNDER on Feb 1st, 2005, 9:17am
I agree with rmsgrey in that equating consciousness with 'soul' requires a quantum leap in both logic and belief.

While (in our case) consciousness => neurons, synapses, neurotransmitters, etc, the converse is not necessarily true.

Any copy of a human being must necessarily involve the mind, and 'soul', by definition, transcends the mind.
Therefore, the 'soul' cannot be copied by the mind.

Title: Re: Revenge of the Right Brain
Post by JocK on Feb 1st, 2005, 3:30pm

on 01/31/05 at 17:56:53, Icarus wrote:
True A.I., if and when it comes, will come from solid programming, not from simply tossing more switches together. To expect that would be similar to expecting a cathedral to appear once you threw enough boards and rocks into a pile. And claiming that because no computer yet has shown sentience, it is strong evidence that A.I. is impossible, is also akin to claiming cathedrals are impossible because piles of rock and board do not turn into them all on their own.


Which implies that so far zero progress has been made towards "true A.I.".

If we all agree to that, the only thing we might disagree on is on how to extrapolate this fact. And yes, that is where gutfeel / judgement / intuition / beliefs etc. enter the picture, and where we can/should agree to disagree. :)

Title: Re: Revenge of the Right Brain
Post by Icarus on Feb 1st, 2005, 3:51pm
I guess it has been too long (about 30 years) since I read that book. My apologies to Heinlein for misrepresenting his story. I know I have read books wherein the idea was that if you add enough computing power, the machine will develop sentience all on its own. I was mistaken in identifying which story, though. Thanks for pointing that out.

Also, rmsgrey, you may have been refering strictly to consciousness, but Jock was refering to something more. I felt that this should be brought out plainly. But then, how are you defining consciousness? How is your consciousness different from what computers experience today in their functions? One possible answer is "self-awareness", but this begs the question of what it means to be "aware". However you define it, you are still at least at the bounds of religion, and generally beyond the bounds of solid science.

towr - sorry to have been unclear. I did not mean anything in particular by "solid" programming. Perhaps you could interpret it as "setting up a neural network with the capacity to become sentient". All I meant is that the main requirement for true A.I. is software, not hardware. The hardware requirements are minimal - just any system capable of acting as a universal computer, and possessing enough memory to handle all the data. Of course, if you want an intelligence you can interact with, then a certain amount of speed is desirable as well.

T&B, the 'soul' does not necessarily extend beyond the mind. Since the soul is not a well-defined concept outside of particular religious views, and those views do not agree in their definition, one cannot be sure that a copy of body would not also copy the soul.

Again, these are questions that cannot be answered at this time. Perhaps with more advancement, the existance of anything beyond the mind can be tested. Then we can argue these things from evidence, rather than speculation.
Title: Re: Revenge of the Right Brain
Post by Grimbal on Feb 1st, 2005, 4:19pm
Just a thought.

Whatever consciousness is, it is real.  It is not something you get "on top" of reality, something that is not measurable by physical means.  I can tell that because else, our consciousness would have no way to affect our actions and we wouldn't hear ourselves say "I am conscious because I know I am".  That is why I think our consciousness is a result purely of our physical brain, and therefore could in theory be reproduced artificially.  Maybe not in a digital computer, though.  And maybe the technology wouldn't allow duplicating a consciousness.
Title: Re: Revenge of the Right Brain
Post by Icarus on Feb 1st, 2005, 4:28pm

on 02/01/05 at 15:30:16, JocK wrote:
Which implies that so far zero progress has been made towards "true A.I.".

If we all agree to that...


Only if we also all agree that you make no progress at all towards a goal until you actually achieve it. I don't know about you, but when I go somewhere, I have to travel through all the places inbetween before I arrive. I personally count reaching these inbetween places as partial progress towards my goal.

No, we have not achieved true A.I., but to claim this means "zero progress" is ridiculous. Neural networks, and the proof-producing program amichail linked to both demonstrate significant progress, I would say.
Title: Re: Revenge of the Right Brain
Post by rmsgrey on Feb 1st, 2005, 4:32pm

on 02/01/05 at 15:51:17, Icarus wrote:
Also, rmsgrey, you may have been refering strictly to consciousness, but Jock was refering to something more. I felt that this should be brought out plainly. But then, how are you defining consciousness? How is your consciousness different from what computers experience today in their functions? One possible answer is "self-awareness", but this begs the question of what it means to be "aware". However you define it, you are still at least at the bounds of religion, and generally beyond the bounds of solid science.

I know I am conscious. I accept that others are conscious as a "least hypothesis" to explain their behaviour. I don't ascribe consciousness to computers largely because they don't interact in a way that would make consciousness a simpler explanation (in my opinion). I don't rule out the possibility that computers are conscious, but if they are, then they aren't telling.
Title: Re: Revenge of the Right Brain
Post by Icarus on Feb 1st, 2005, 4:41pm
But what do you mean by "conscious"? It does not explain the behaviour of others or prove useful or not useful in explaining the behaviour of computers until it has some sort of working definition.
Title: Re: Revenge of the Right Brain
Post by rmsgrey on Feb 1st, 2005, 7:59pm

on 02/01/05 at 16:41:21, Icarus wrote:
But what do you mean by "conscious"? It does not explain the behaviour of others or prove useful or not useful in explaining the behaviour of computers until it has some sort of working definition.

I know exactly what I mean by "conscious" in this context (or, at least, I can tell whether any given proposed usage is appropriate) but I'm not sure I can define it properly.

A reasonable stab would be: "aware of oneself as a self-aware self." or "Being an 'I'"
Title: Re: Revenge of the Right Brain
Post by towr on Feb 2nd, 2005, 12:38am

on 02/01/05 at 19:59:53, rmsgrey wrote:
A reasonable stab would be: "aware of oneself as a self-aware self."
Isn't this similar to "conscious of oneself as a self-conscious self."?
Consciousness/awareness is a rather big philosophical question..

imo, at least part of it should be reflection on ones own thoughts/reasoning and actions. "What am I doing, and why?"
But it's hard to go beyond that without begging the question.
Of course even the 'why' is already asking something peculiar, it asks for motivation. Is there really anything we want, or is it just some physical aspect of us giving us the illusion we want something.
Title: Re: Revenge of the Right Brain
Post by JocK on Feb 2nd, 2005, 3:25pm

on 02/01/05 at 16:28:27, Icarus wrote:
Only if we also all agree that you make no progress at all towards a goal until you actually achieve it. I don't know about you, but when I go somewhere, I have to travel through all the places inbetween before I arrive. I personally count reaching these inbetween places as partial progress towards my goal.


Are you always drunk when you attempt to "go somewhere"? ;) What I mean is that indeed I don't classify a random walk (that is unlikely to lead you to "inbetween places") as partial progress.

You made the analogy between the current state of A.I. and 'progress' resulting from random activities: "claiming that because no computer yet has shown sentience, it is strong evidence that A.I. is impossible, is also akin to claiming cathedrals are impossible because piles of rock and board do not turn into them all on their own."

I liked your comparison between AI-research (as an attempt to reach "true A.I.") and random walks. (It now seems you didn't intend to make that comparison.) I truly haven't seen any AI-results so far that demonstrate even infinitesimal progress towards the distant goal of "true A.I.".

Perhaps we have different opinions about what is meant by "true-A.I."? (Your mentioning of neural networks and proof producing programs as examples seems to suggest so.)
Title: Re: Revenge of the Right Brain
Post by Icarus on Feb 2nd, 2005, 4:27pm

on 02/01/05 at 19:59:53, rmsgrey wrote:
I know exactly what I mean by "conscious" in this context (or, at least, I can tell whether any given proposed usage is appropriate) but I'm not sure I can define it properly.


If you can't define it, then I have to question just how "exactly" you know the meaning.


Quote:
A reasonable stab would be: "aware of oneself as a self-aware self." or "Being an 'I'"


Which brings up the question of what "aware" means. One can make strong arguments that most robots demonstrate some self-awareness in their actions. They act in ways that indicate a knowledge of their position in relation to their environment. Of course, this is the result of "simple" programming that fall far from what we think of as consciousness. But how does our own ego differ from this?

Jock - "all the places inbetween" is all the places along the path taken from one point to another. This does not require any randomness in the choice of path. To claim that neural networks, learning systems, proof-generating routines, etc. represent infinitesimal progress requires some reliable estimate on how far we have to go, and that this estimate is vastly beyond these systems. But the best research to date on intelligence seems to indicate the opposite: Intelligence is made up of interactions between very primative subsystems. We are reaching the point where we can simulate the primative subsystems. That is a huge step towards A.I.
Title: Re: Revenge of the Right Brain
Post by Noke Lieu on Feb 2nd, 2005, 4:50pm

on 02/02/05 at 16:27:55, Icarus wrote:
If you can't define it, then I have to question just how "exactly" you know the meaning.


Love? Okay... dictionary.com goes for
n.  A deep, tender, ineffable feeling of affection and solicitude (erm, shouldn't hat be solace?) toward a person, such as that arising from kinship, recognition of attractive qualities, or a sense of underlying oneness.

actually, not bad as it goes. Was hoping to with an argument along the lines of "you don't need to know what it is to feel it." I will let you all make that argument for me in your own heads.
Title: Re: Revenge of the Right Brain
Post by rmsgrey on Feb 2nd, 2005, 8:01pm

on 02/02/05 at 16:27:55, Icarus wrote:
If you can't define it, then I have to question just how "exactly" you know the meaning.

I have an implicit definition that (I believe) enables me to differentiate between things to which the word applies, and things to which it doesn't. There may well be a grey area, but the same would be true of any explicit definition (arising from ambiguities on the words used for the definition). After all, most of our vocabulary is derived from an inductive process rather than by having dictionary definitions read to us, but few people would quibble if I claimed to know exactly what the word "apple" means.

Quote:
Which brings up the question of what "aware" means. One can make strong arguments that most robots demonstrate some self-awareness in their actions. They act in ways that indicate a knowledge of their position in relation to their environment. Of course, this is the result of "simple" programming that fall far from what we think of as consciousness. But how does our own ego differ from this?

The major difference is that the robots are not (so far as we know) aware of their awareness. That introspective quality is vital (and the main reason my first "definition" is phrased as it is).

Quote:
Jock - "all the places inbetween" is all the places along the path taken from one point to another. This does not require any randomness in the choice of path. To claim that neural networks, learning systems, proof-generating routines, etc. represent infinitesimal progress requires some reliable estimate on how far we have to go, and that this estimate is vastly beyond these systems. But the best research to date on intelligence seems to indicate the opposite: Intelligence is made up of interactions between very primative subsystems. We are reaching the point where we can simulate the primative subsystems. That is a huge step towards A.I.

A better metaphor for AI research than random walks is probably that of a maze - you know how to recognise the goal, and know roughly which direction it's in, but have no idea which routes lead you there, which lead to dead ends, and which lead to somewhere entirely unexpected, but the closer you are to the goal, the more useful even the dead ends are because they form the spaces around the path(s) to the goal. And it's very hard to judge progress without the benefit of hindsight since the path-distance to the goal is often very different from the straight-line distance...
Title: Re: Revenge of the Right Brain
Post by Sir Col on Mar 1st, 2005, 4:24pm
In light of the discussions that took place in this thread about the limit of computers being able to prove results, you may find this link very interesting:
http://www.cse.iitk.ac.in/~amit/courses/768/00/mayanka/
Title: Re: Revenge of the Right Brain
Post by rmsgrey on Mar 1st, 2005, 6:22pm

on 03/01/05 at 16:24:57, Sir Col wrote:
In light of the discussions that took place in this thread about the limit of computers being able to prove results, you may find this link very interesting:
http://www.cse.iitk.ac.in/~amit/courses/768/00/mayanka/

An interesting consequence of the Church-Turing Thesis ("all effective models of computation are equivalent") is that a sufficiently complex railway layout (with purely mechanical automatic points, no computerisation) can, in principle, solve any computable problem that a human can.

If AI is possible, then, in principle, a train network could become conscious (before taking this as evidence that AI is impossible, consider that this would require a very large, and probably deliberately designed, network, so is no less plausible than AI in any other computer).

I've had a sneaking suspicion for a while that most of mathematics is computable - which Sir Col's link supports. Of course, to the committed AI disbeliever, the portions concerning choosing what to prove and finding new methods of proof are a yawning chasm compared to the small steps of applying known methods to chosen problems. Equally, for a true believer, creating solutions (creatively) is most of the battle, and expanding the field to wider problems a mere detail.


Powered by YaBB 1 Gold - SP 1.4!
Forum software copyright © 2000-2004 Yet another Bulletin Board