a few examples:
An artificial intelligence draws a picture identical to hello kitty, without having ever seen hello kitty before.
an algorithm for generating science fiction novels creates a work nearly identical to some already published work, again without ever having received any of the copyrighted information as input.
The reason I ask is because it is likely that this will eventually happen, and the spirit of a copyright is that it is intended to prevent someone from benefiting from another person's legitimate work. These cases strike me as a bit odd primarily because (given sufficient auditing of the generating system) it can be proven that the work (although nearly identical) is also original.