After having (successfully) resorted to Australian justice to overcome the resistance of that country’s IPO, which refused to grant patent rights to inventions generated by an artificial intelligence algorithm, scientist Stephen Thaler intends to do the same in the USA, a country whose IPO has also refused such attribution. https://news.bloomberglaw.com/ip-law/inventor-robots-deserve-patent-rights-federal-circuit-to-hear
It seems clear, at least to me, that an algorithm with inventing skills but that has learned to do so with substantial human input, that the invention rights belong to its human creator, but it is difficult to extend this reasoning to algorithms that teach themselves “an algorithm based solely on reinforcement learning, without human data, guidance or domain knowledge”