In the new technological era, artificial intelligence (AI) reigns supreme. With the assistance of AI systems, society is undergoing a radical transformation. AI may not only soon replace human labor in many industrial sectors, but as AI gains the power to generate greater inventions, it may also outsmart human inventors. How should patent law and policy adapt to the formidable challenges of the AI era? One of these challenges, addressed by patent offices and courts in 2020 and beyond, is whether AI inventorship should be recognized. The United States Patent and Trademark Office and European Patent Office declined to recognize the autonomous AI system DABUS as an inventor despite its two inventions. Courts in the United States and United Kingdom upheld these rulings. However, the Federal Court of Australia and the South African Patent Office steered patent law in the opposite direction, accepting DABUS as an inventor and thereby legally recognizing Al inventorship. This Article argues that these divergent approaches to determining the legal status of AI inventorship fail to address proper policy considerations central to shaping AI and patent law in service of the public interest. Applying broad-based, forward-looking policy considerations, this Article puts forward three legal principles for protecting ALgenerated inventions. The first principle draws on the doctrine of "piercing the corporate veil" to ascertain the sole patent proprietor of AI-generated inventions. It attempts to remove the unnecessary cost of protecting AI systems that are incapable of securing ownership of their inventions. The second principle considers the capacity to take legal responsibility as a means of evaluating whether AI systems should be recognized as inventors. It channels an ethos mandating that any grant of patent rights be conditioned on certain legal responsibilities. The third principle dictates that patent protection of AI-generated inventions must promote robustness of the public domain through the free flow of information and knowledge not subject to proprietary control. Together, these principles can better protect a wide range of public interests implicated in the patent protection of AI inventions.
Artificial Intelligence Inventions,
50 Fla. St. U. L. Rev.