“The robots develop their own musical culture. There are no pre-programmed musical rules.”
Posted: November 3rd, 2008, by Stan TontasNew Scientist report on experiments with music & artificial intelligence. One robot “sings” a few notes to the other, which sings back. If the 1st robot thinks the songs are similar, they agree to remember the sequence.
The researcher hopes that this will lead to the robots helping “him to compose music that no human would ever come up with”. I am sceptical, not just because the attached video (Quicktime format, <1 minute) is pretty underwhelming.
If the robots choose the sounds which are the same, where is the scope for development? Can robots do improv? Or will this just lead to them adopting a fixed repertoire (which, to be fair would be pretty cool if they started from zero with regards to the robots making sounds)?
Surely the thing that would lead to new sounds would be the criteria that the robots use to reject or adopt a “song” – and here it seems to be “this is the same as what my mate sings”. That brings you a zillion indie bands, not “new musical cultures”. I get that there are “emergent” properties from simple systems (think fractals) but the way this story is described, these are being selected out of the robots. If the song isn’t sung back to you it’s discarded.
Poor bastards! Not even self-aware but still subject to merciless criticism by its only friend! Imagine a robot Jimi Hendrix. No-one singing your tunes back to you. Cursed to artistic isolation until musical tastes catch up to you. Singing unheard till your batteries run out or you break down through overwork in the face of an unappreciative world, found disassembled in a bath, OD’ing on acid and lithium…
Maybe the answers are in the full research paper.
November 12th, 2008 at 9:07 pm
DESTROY ALL ROBOTS!
Except friendly ones.