Human to machine innovation: Does legal personhood and inventorship threshold offer any leeway?

Abstract Artificial Intelligence (AI) continues to be a powerful tool in the research and development ecosystem. AI computers are invented to assist human invention and also created to invent. Where an AI is created to invent, through self‐learning, they can interact with set of data presumably created by humans and as a result, a new patentable invention(s) can emerge. However, where the AI inventors and the resulting inventions sit within the inventorship legal framework, and the theory of legal personhood continues to raise legal and policy questions that challenge some underlying or presumed settled intellectual property law assumptions. One of the contentions has been the implications of the AI machine's autonomous inventions on the legislative and judicially established threshold for patent inventorship and the jurisprudential theory of legal personhood. The judicial decisions in the United States of America (USA), United Kingdom (UK), and Australia in the Device for the Autonomous Bootstrapping of Unified Sentience (DABUS) patent applications have given judicial certainty on whether AI machine inventors qualify as inventors. However, they also reawakened the debate about the need to sustain patent incentives for AI innovations. This article draws from the inventorship threshold in the UK and US following the court decisions in the DABUS cases. The judicial decisions of courts and the administrative judgements of national Intellectual Property Offices (IPOs) relating to inventorship as well as the theory of legal personhood, reveal that an AI machine invention can be patent eligible. However, the machine does not satisfy the inventorship criteria and consequently is incapable of being named an inventor. On the other hand, the inventorship requirement of contemporaneous conception and reduction to practice meant that an AI owner/programmer may not satisfy the requirement of inventorship, even though he/she programmed the inventing machine. These decisions and judgements favour an implied situation where autonomous AI inventions could be without named inventors and owners. Consequently, those inventions will automatically form part of prior arts thereby rendering myriads of future human and AI inventions obvious or already existing in the public domain. In contributing to the discourse, this article advances the argument that to optimise the patent system, national IPOs and the courts can rely on ‘simultaneous conception and reduction to practice’ to recognise the programmer/owner or other relevant stakeholders in AI innovation as the inventor of AI autonomous inventions.