Mike, Your argument is Fallacious. There are many human musicians and music producers that "create" music within a particular genre. Country, Hip-hop, Pop, etc. These musicians are making "new" music within a particular genre and are very comfortable doing it, and very lucrative as well. Then there are musicanswhich combine genres as well. These are all variations, within a genre, and across genres. I think you have a basic meme running through your brain that says "Computers can't be creative" and thataxiom is a the core of your inference processes. You should extricate (or suspend) that premise if we're allgoing to get anywhere. No one has to show you anything, it is you that must adapt to the reality of the world. The world model you'vementally constructed is always in error, and must be adapted to the evidence that is all around you but which you cannot percive. "A system of assimilation tends to feed itself." ~ J.Piaget This means you accept what youare comfortable accepting and reject what you are used to rejecting. But it is you that must shift your biases if you want to be truly creative and constructive. Throw away your old patterns of thought, your old assumptionsand try new premises for a change. ~PM Date: Fri, 29 Nov 2013 12:47:14 +0000 Subject: Re: [agi] Composing music and other creative exercises From: [email protected] To: [email protected]
Steve, This is yet another example of AGI-ers delusions. Here's a music program -music is creative - therefore the program is creative. It's not creative at all. It's easier to think about this if we start with visual arts programs. By your logic a "Mondrian" program - or we could equally cvonstruct a Pollock inblot program - is creative, because it gives you Mondrian variations. Neither are/would be creative. They are "arts Lego kits" - sets of basic abstract shapes, basic rectangles or inkblot lines for example - on which they construct variations. And that's it. They can't add/create one new shape. Period. They're just recipe variations. OLD recipes, OLD [MondrianPollock] paintings. Nothing new here. Ditto music programs. "Improvisation" programs similarly construct a set of variations on a "music Lego kit" - a set of basic musical notes, chords, refrains, whatever. And that's it. They can't add/create one new note, noise, instrument. Period. They're just recipe variations. The use of random numbers makes only a trivial and no real difference. OLD music. OLD C & W, rock, classical etc music. If they were creative, they would function like human composers - it/musical AGI would be a WHOLE DIFFERENT KIND AND CULTURE OF PRODUCTION - a different kind of intelligent, productive activity. With rational, narrow AI you start with a fully specified formula/algo and produce something old. With human composers, you start with a brief (or they brief themselves) - "give me a rap song like Kanye's Bound about infidelity, but with monastic choral music instead, something like that..." Or "here's a nice refrain/chord - see what you can do with that..." And you produce something NEW, not old - even if the newness is at times only a slightly different stew, collage - strictly "incremental" as opposed to "transformational"" creativity. In the arts, - - you start with some form of IDEA/brief - ALWAYS - not a complete step-by-step formula-algo - (an algo for an algo). [And THIS IS EQUALLY TRUE OF COMPUTER PROGRAMMING as distinct from finished programs - and scientific and technological creativity. The creation of new algos always starts from ideas not other algos] That's the Woz Test - an AGI robot must be able to start with GO TO THE KITCHEN - an *****idea/brief*** - not any kind of formula/algo. Just the briefest outline. And then the robot will have to create a new journey forged as it goes along in search of this new kitchen in this new house, rather than reproducing a precise variation on some old journey, as a current factory robot would Neither you nor anyone else gets this - and I need to expand on it much more fully. Creativity is a WHOLE DIFFERENT KIND, LEVEL AND CULTURE OF PRODUCTION - smart, high-level intelligence as opposed to the dumb, low-level intelligence of algos./routines. AGI requires a 2nd computer revolution - Turing introducing the rational, formulaic/algo process was the first. The second is the introduction of the IDEA-based machine/computer project. - Project not process. An adventure into new territory, not a foregone conclusion of a journey in old territory. When you tell a real AGI robot, as you do with a human, - FIND THE KEY IN THAT ROOM, PACK MY CASE, CLEAR THE ROOM, FIND BEN IN THAT CROWD, MAKE COFFEE IN THE KITCHEN. you arre giving it a creative, outline brief and it has to work out the details of that brief for itself, and come up with something - a journey - which will be new, even if only incrementally new as distinct from a transformational new work of art. What are you doing, Steve, like every other AGI-er, when confronted with the unanswerable challenge - SHOW ME A SINGLE ALGO THAT DOES OR COULD PRODUCE A SINGLE NEW ELEMENT is respond: "but algos are creative, aren't they, somehow, somewhere - they must be - please God let them be creative, because they're all I know..." No they're not - they're totally rational, totally dumb, totally "old". AGI is going to be a computing REVOLUTION - the biggest thing since, perhaps even bigger than, Turing. "Smart" computers/robots as opposed to the "dumb" computers/robots we have at the moment. Computers with IDEAS that can CREATE new courses of actions on their own, as opposed to computers with algos that can only iterate old courses of action, predesigned for them by human programmers. Independent machines not puppet machines. On 28 November 2013 22:31, Steve Richfield <[email protected]> wrote: Mike, et al, In the distant past I have worked with creative composers to create two very different programs to compose music. The logic of these programs was more in deciding what NOT to do than what TO do, so there was generous use of a random number generator, followed by logic that rejected most selections. A common situational challenge was that there was no acceptable next note, so time to back up or start over. While this fit the "programmed" model you so like to reject, it ALSO reflected the mindset of most composers. Sure there is an occasional maverick who deviates from one of the many patterns, and in so doing creates a new pattern, like switching between a major and a minor key in mid-piece. However, people like these are in the EXTREME minority - about as rare as malfunctioning computers, so you could run less creative programs on many computers, and sometimes be surprised over what a malfunction might bring. For a good discussion of these deviations, you might watch the now-unfolding story aboutf the lawsuits over the piece Blurred Lines, which is a highly creative piece that borrows from another piece, but in ways that are so subtle as to probably NOT violate (present) copyright laws. http://www.youtube.com/watch?v=yyDUC1LUXSU Apparently, creative music CAN be composed by an expert system designed to do that. The amazingly simple rules for such systems come from centuries of creative composers. Such a computer would probably NOT create these deviations, but then again, neither do most composers. It appears that creativity comes at more than one level. A computer might be able to solve all equations that people can now solve, but never push back that frontier to solve equations that people can NOT now solve. Similarly, a computer might be able to create music as good as a graduate from a major music school, but never create the likes of Blurred Lines. without something else first pointing in that direction, which is what the lawsuits are all about. Robin Thicke readily admits that he was actually listening to Marvin Gaye's music as he was composing Blurred Lines,but claims that Blurred Lines is NEW in ways that do NOT tread on copyrights. My conclusion is that computers can now already be creative, but there are limitations that apply equally to most people. We CAN now program great skill, but not yet program deviant genius. Any thoughts? Steve AGI | Archives | Modify Your Subscription AGI | Archives | Modify Your Subscription ------------------------------------------- AGI Archives: https://www.listbox.com/member/archive/303/=now RSS Feed: https://www.listbox.com/member/archive/rss/303/21088071-f452e424 Modify Your Subscription: https://www.listbox.com/member/?member_id=21088071&id_secret=21088071-58d57657 Powered by Listbox: http://www.listbox.com
