Meat Machine

Okay, so this is not the argument I thought I would get into, but off and on for the past few weeks I've found myself discussing the nature of the human brain with a computer scientist. Today, he made me listen to Beethoven and then proceeded to argue that no computer could ever come up with something so complicated. Not just that humans are better at it, or that modern computers don't have enough power or the right programming to compose a symphony, but that they fundamentally cannot. The brain is not a "meat machine", he insists.

My argument is that it is clearly possible for some physical device to create art, because clearly humans can, and that computers still follow the same laws of nature, just differently. He pretty much ignored this, and pointed back to Beethoven, emphasizing that it is so complicated, and look at how he put these parts in to have this effect and that it was such a bold movement and couldn't possibly be just a new step in cultural evolution. I told him that's not an argument at all, and it's no better than a creationist pointing at the human eye, but he guffawed and pointed back to Beethoven.

Now, it may well be that he is, in some sense, right, or that I'm misunderstanding him (I don't know enough about how neurons make connections or how they could be modeled in another medium, particularly silicon) but the thing that really struck me is that he's trying to make a philosophical claim that there is some human essence that is not contained in the brain: a Cartesian Soul, although he again refuses to address it. The man has a P.h.D in mathematics and studies biological systems modeling, so I'm simply baffled by this.

Does anyone have any resources or opinions on this? I'll probably see him again on Thursday. Rhinoq? Scathach? Judge?

Disclaimer: I know jack shit about the science of this matter, just the logic.

It seems to me that his only pseudo-logical argument agaist a computer having the capacity to produce art is that art is simply too complicated. Is he somehow suggesting that computers are at their maximum level of complexity, or that there exists such a boundary, or that the human brain is well above this boundary? As far as I care, the brain is a computer; so there is clear proof that computers are capable of art.

[Rhinoqulous]

Your friend seems to assume that the "beauty" or "art" of Beet-hoven's music exists in the music itself, and not from his conscious experience of the music. Have him listen to Autechre (damn-it, won't link to the song; just search for Autechre on the BLEEP site, and take a listen to the song Gantz_Graf for this reference) and see if he considers it to be "art" or "beautiful". If he doesn't think it is a "beautiful" song, I can fill pages on why that song blows Beethoven out of the water (well, maybe not Beet-hoven, he's da bomb, but definitely that chump Mozart). The point is it doesn't matter how complex the system is, the individual imposes the “art” of the system/object on the object. Run to the library and find a book of post-modern art or found-art, and see how art can come from the description alone of the object, regardless of the complexity or origin of the object.

So yes, a machine can produce "art". Whether or not that same machine can "appreciate" said art is a different question entirely.

[Choobus]

Imagine if computers did become sentient. That would make wanking off to internet porn really awkward.

[Rhinoqulous]

Choobus wrote

Imagine if computers did become sentient. That would make wanking off to internet porn really awkward.

:lol::lol::lol::lol::lol:

You better not get any in my keyboard this time, Dave.
:lol:

whoneed wrote

Today, he made me listen to Beethoven and then proceeded to argue that no computer could ever come up with something so complicated.

Point him to DNA computers.

Rhinoqulous wrote

Have him listen to Autechre (damn-it, won't link to the song; just search for Autechre on the BLEEP site, and take a listen to the song Gantz_Graf for this reference) and see if he considers it to be "art" or "beautiful". If he doesn't think it is a "beautiful" song, I can fill pages on why that song blows Beethoven out of the water (well, maybe not Beet-hoven, he's da bomb, but definitely that chump Mozart).

So that is computer generated? Do they have any more that you know are 100% computer generated? Something tells me that wouldn't convince him.

Rhinoqulous wrote

The point is it doesn't matter how complex the system is, the individual imposes the “art” of the system/object on the object. Run to the library and find a book of post-modern art or found-art, and see how art can come from the description alone of the object, regardless of the complexity or origin of the object.

So yes, a machine can produce "art". Whether or not that same machine can "appreciate" said art is a different question entirely.

Yeah, he said a lot about how a computer couldn't design a piece of art to have an effect on people, so I assume he meant more that a machine cannot appreciate it. My argument is that the brain is a computer, programmed by evolution to find and appreciate patterns, and the fact that we can describe our feelings on the matter in no way means that we are not programmed, or that we can't, potentially, build a machine that does exactly what our brain does.

[Mog]

I'd have to agree with Whoneedsscience on that Autechre piece. That is not convincing. If I was to attempt to make a computer that will generate complex artistic sounding music that is pleasing to the human ear, I'd probably build a database of popular melodies and create a program that derives music roughly out of those melodies. I suspect that many so called creative humans do something like that subconsciously, anyway. I'm not a musician, but I can recognize thousands of popular songs within 5 seconds of hearing it. So, I think the real question of study is why do we perceive some things as being beautiful and other things as being ugly? If we can crack this, then we can insert the proper amount of prejudice to the randomly generated melodies of computers.

[Eva]

...huh huh......meat machine....huh huh huh.....

sounds like the title for a blaxsploitation (sp?) porn flick......

[Rhinoqulous]

Mog, Who, you seem to have missed my point. I used the Autechre song as it's not "traditional" music (I'm sorry that I made it appear that it was completely computer generated, it's not; it's "non-repetitive music, where each measure is distinct from every other measure in the song), something most people wouldn't find "aesthetic pleasure" in. My point was more about "art" than AI. A computer can generate art because pretty much anything can be "defined" as art.

As for a computer intentionally generating art, it seems odd that your friend would insist that it can never happen. His insistence that our brains are not "meat machines" is absurd. If he wants to say current computers are nothing like the human brain, well, of course they're not. By why he thinks that it would not be logically possible to generate something equable to human consciousness (and all that would imply) from something other than the human brain is narrow-minded.

[a different tim]

Turing dealt with most of these claims in his original paper proposing AI (and also disposes of the entire Penrose "computer's can't be sentient because they can't do Godel questions" book in one economical sentence, 40 years before Penrose wrote it.....).

Well worth reading. I linked to it before in one of the free will arguments but it's one of the posts that got lost.

[Tenspace]

Your friend needs to read Deutsch.

Rhinoqulous wrote

Your friend seems to assume that the "beauty" or "art" of Beet-hoven's music exists in the music itself, and not from his conscious experience of the music.

Touche. I'm reminded of things like the pomo generator which produces essays which are disturbingly like the "real things".

And what about Wave DNA which in the words of a local geek blogger: "is a "music reverse engineering tool" whose purpose is to analyse music for patterns. It breaks down music into fundamental units of pitch, duration and "feel" and perfoms analysis on those elements and their arrangement. The potential uses for this application are vast and interesting, ranging from analysing what makes a good commercial jingle (by analysing the most-remembered ones and seeing what their common qualities are) to a Ph.D. thesis project in which a researcher is trying to determine what makes a song a lullaby. "

Doesnt' seem too big a stretch from that to computer-generated music which is indistinguishable from that created by humans.

Tenspace wrote

Your friend needs to read Deutsch.

this?

[a different tim]

Or possibly this.

Wolfram tones. Good for a laugh.....the "signalling" ones can turn out quite haunting. Eno has also done a fair amount of generative music. You can even go back to Reich and co using physical processes to make music. I do "pendulum music" with my music tech students every year just to try and break them out of the idea that they have to control every aspect of music making.....

On the other hand, Sloboda (Cognitive Psychology of Music, 1985) gives some examples of music generated by Sundberg and Lindblom according to a generative schema drawing on Chomsky, and you can tell from the score, without even listening (and I'm not a fluent score reader) which ones are generated and which ones aren't. Mind you that is a pretty old example and you could argue that there is something wrong with the schema, or something hard to duplicate about the genre (Swedish nursery tunes!) instead of it being a problem with generative processes per se. Ah, yes, it's all coming back to me now. It's a while since I looked at that stuff.

I rather like generative stuff myself although I tend to prefer stuff like Reich phase pieces (which I think are related as they again use physical processes - in this case tape speed - to generate novel melodies). In fact I really like Reich phase pieces. They're great. I urge you all to listen to lots of Steve Reich. When I were a lad we didn't have computers etc etc.....