On Wed, 2004-11-10 at 06:21, Amy Alexander wrote:
> i think it illustrates two ways of looking at the idea of an "algorithm."
> their idea is the more traditional research-oriented idea: can an
> algorithm resemble/replace a thinking human? toplap's idea is that we're the
> humans and algorithms are something that humans "do", like making
> music or making images.
> they: algorithms =~ humans ... we: algorithms =~ musical instruments.
> and/or
> they: composition-related interest in algorithms
> we: performance-related interest in algorithms
I think livecoding is a very pragmatic and in your face idea, it seems
quite punk rock to generative art's prog rock.
or something...
> as to dave's question about AARON: that was the idea... how much like an
> independently painting human could AARON become? the interest was in how,
> over the years, cohen could give it increasingly less information about
> what or how to paint, and it could make increasingly independent
> "subjective" decisions. but what i think is most interesting was that of
> course AARON's paintings exhibited cohen's aesthetics... so the
> independence with which it could paint was juxtaposed by its strong
> dependence on obviously very subjective algorithms written by its
> author. AARON's success was that cohen didn't just program it - he
> brainwashed it, which i like very much as it brought out that very
> important subjective aspect to AI...
I guess it make a bit more sense in that context. However, intelligence
(or the appearence of it) comes from something reacting to the world,
and I think it's hard to claim AARON has much to say about AI when it's
so linear - it would be a much more interesting experiment to make an AI
art critic, I'm sure it's out there somewhere :)
> although the human performers are improvising with the computer, the
> computer doesn't really "know" it's improvising live... it knows that
> it's accepting input in real-time, but it doesn't care about the
> "liveness" of its output. unless it's a very complex algorithm, a computer
> will output the same thing whether you give it 5 seconds or a week to
> think about it - i'm not sure the word "improvise" can really be applied
> to a computer... this is where the toplap "human" approach is much different
> - it sure as heck makes a difference that we're "outputting" live...
a computer will always output the same thing given the same input, no
matter how complex the algorithm, as it's entirely, utterly, boringly
deterministic. thats why we're so odd. or are our inputs just much more
complicated? etc etc...
whatever, there seems to be a range of possibilities from banging two
rocks together to make a song, all the way to worrying about building an
entire brain that knows how to bangs rocks together to make a song, and
we're somewhere in the middle :)
dave
Received on Thu Nov 11 2004 - 02:05:27 GMT