Abstract: Artificial Intelligence (AI) systems are increasingly being used in various fields to generate novel and creative inventions. However, the patent law system has not kept pace with these innovations and consequently there are several unresolved issues with respect to the patentability of inventions generated by AI systems. This essay focuses on two of these issues: the grant of inventorship status to AI systems and the application of the current non-obviousness standard to AI generated inventions. It begins by discussing the developments and the state of the art in the field of invention generating AI systems. It then goes on to analyze the inventorship requirement and examines whether AI systems qualify as ‘inventors’ under the existing patent regime. In this regard it examines the procedural and technical requirements for filing patent applications, the jurisprudential understanding of an ‘inventor’, and the decisions of the few Patent Offices which have dealt with this issue. It cautions against the dismissal of AI patent applications on purely formalistic grounds when the AI systems in question fulfil the substantive requirements for inventorship. It further examines the implications of such dismissals and makes policy recommendations. It then examines the current non-obviousness standard for patentability and argues that the current standard is inadequate for dealing with AI-generated inventions. It concludes by proposing a new standard which substitutes the human PHOSITA with an equivalent skilled AI system, focusing on reproducibility rather than obviousness.
Authors: Arvind Sankar is a 5th Year student at the National Law University, Odisha. He may be contacted at email@example.com.
Jitmanyu Satpathi is a 5th Year student at the National Law University, Odisha. He may be contacted at firstname.lastname@example.org.
The full text of the article can be found as a PDF here.