It is now possible for AI systems to generate novel inventions without meaningful human direction and control. Should such inventions be patented? The prevailing consensus, confirmed in recent test cases and official guidance, is that patent law only covers inventions by natural persons (i.e., humans). This, however, sometimes creates an odd situation in which AI-generated inventions cannot be patented, nor can the humans responsible for those systems gain patent rights indirectly through the op…
Read moreIt is now possible for AI systems to generate novel inventions without meaningful human direction and control. Should such inventions be patented? The prevailing consensus, confirmed in recent test cases and official guidance, is that patent law only covers inventions by natural persons (i.e., humans). This, however, sometimes creates an odd situation in which AI-generated inventions cannot be patented, nor can the humans responsible for those systems gain patent rights indirectly through the operation of the law. In this article, we argue against this prevailing consensus. We present five reasons for thinking that AI-generated inventions should be patentable and that AI systems should be legally recognized as inventors. In making this argument, we do not claim that modern AI systems have acquired some significant legal or moral status that is equivalent to humans. Our argument is more practical in nature. We argue that failing to recognize AI inventorship will have negative repercussions for economic development and innovation, at a time when AI assistance is needed.
