@sarahzedig the ruling wasn’t so much about the origin of the datasets but that only work created by a machine cannot have a copyright

i think that’s a far more damning statement, and it’s also a crucial thing for talking about. the issue isn’t just one of the training sets, it’s that can a work generated truly by a machine be a creative one, and the USCO currently holds the opinion of “no”

@sarahzedig i liked this video up until it started talking about legal stuff in a way which was just … false (or at least, lacked critical nuance)

e.g, he seemed to make the assumption the reason that AI comic cannot be copyrighted was because it used work that was unlicensed, so large corps could use this tech and be fine, as long as they trained it on their own stuff