Grammy for AI generated music

One aspect of interest (at least to me) is that machine generated output is not copyrightable. And since the music publishing and sound recording business depends on copyright to control compositions and recordings, I'd be curious to know how the originator (and whoever claims ownership) parses the machine generated "authorship" from that created by humans. Conceivably, there could be a copyright in the sound recording if some human is involved in the performance, production and/or engineering elements. This whole area is not new in copyright doctrine but AI has demanded that these fundamental principles be revisited; it was the subject of a recent study by the Copyright Office, which prompted some political action (which I do not want to delve into). 

New technology always challenges old paradigms. Yes, to get a new voice, it trains on existing voices. One of the open questions.

@ghdprentice - yes a paradigm challenge, but the issue you raise is on the "ingestion" side- taking copyrighted material of others to "train" the platform. That is being litigated as well. But the issue I'm raising about the resulting "product" (the example posed in this thread) is about control of that product, which equals money. If there is no copyright in the resulting work, the purported owner cannot stop others from exploiting it with impunity. And that's a big issue for music publishers and sound recording owners, who make their money by licensing due to copyright; (the alternative-- unlicensed use-- being infringement). 

AI doesn’t create art… without its human intelligence repository it is inert. 
 

AI music is just an algorithmic rearranging of a library of other intellectual property. 
 

:/