Patenting Inventions Generated by Artificial Intelligence: The Way Forward

Abstract

In the last two decades, we have witnessed significant advancement in the development of Artificial Intelligence (AI) which has opened up unprecedented frontiers, accelerated disruptive innovations, solved complex problems, and reduced human efforts, in almost every industry. While the impact of AI on science and technology is still unfolding, there are currently cases where AI - with the aid of techniques such as machine learning, deep learning and neural networks - have independently generated inventions and have been named as inventors in patent applications. Ordinarily, these inventions should qualify for patent protection. However, the current patent laws world over only contemplate natural persons as inventors. This article will examine whether AI-generated inventions will be able to satisfy the fundamental patent requirements of ownership, non-obviousness, patentable subject matter and disclosure, and the difficulties patent offices will and are facing with the advent of AI-generated inventions. Finally, the article will proffer a way forward on patenting AI-generated inventions by suggesting that the developer or programmer of the machine be recognized as the inventor, the prior art should be the field of technology where the AI-generated invention is connected with, the PHOSITA should be a trained AI model, and a detail description of the relevant training data and algorithms should be sufficient disclosure.