On the Patentability of Artificial Intelligence:a review of recent cases from the U.S.

Abstract

As new developments in the field such as neural networks and machine learning have transformed the expectations of AI technology and what it may be capable of, the importance of AI grows exponentially not only in industrial and commercial spheres but also in educational, social and even cultural spheres of our society. Given the importance of AI technology, it follows that the inventors and those involved in the commercialization of the technology would want to protect and exploit the economic value of the technology to its full extent. Obtaining a patent is still one of the most fundamental tools in protecting the economic value of an invention by reserving its exclusive rights to the owner of the patent. However, due to the recent tightening of the requirements in the field of patent law regarding the question of patent-eligibility, it is not entirely clear whether and to what extent AI technology could be patentable. The decision by the U.S. Supreme Court in Alice v CLS Bank and subsequent decisions by the Court of Appeals for Federal Circuit (CAFC) have cast doubt on whether patents that have already been issued for AI satisfy the new tests for patent eligibility and it is difficult to predict with any certainty whether new patent applications in these technologies could satisfy the requirements for patent eligibility due to the fact that these requirements are still in the process of evolution. The traditionally accepted incentive theory would indicate that a strong patent system is the foundation for strong innovation and provides motivation to inventors and offers a known mechanism to evaluate potential returns to potential investors. Although the recent cases have introduced a real element of uncertainty into the system, this will eventually diminish with an accrual of case law through evolving court decisions, and even perhaps new legislation. In the meantime, however, inventors and investors still need to be able to protect their AI technology with patents, even though patents may not necessarily offer a foolproof or ironclad protection they may once seemed to hold out, while knowing that the system was never a guarantee of any value in the invention.