The relationship between Patents and Artificial Intelligence (AI) is of paramount interest to academics and scientists alike. As discussed in our previous newsletter, various organisations and Intellectual Property (IP) offices across the world had called for recommendations and views regarding legal personhood to AI, as well as the impact of recognising AI, on the patent and inventorship ecosystem. The United Kingdom Intellectual Property Office (UK IPO) has received responses from various quarters and this week we look at the submission by ACT-The App Association.
The App Association is a conglomerate of about 5,000 small development companies that deal in business software applications, operating across the mobile economy. Its members play a critical role in developing new products across consumer and enterprise use cases, enabling the rise of the internet of things (IoT). Software applications are the result of the concerted and dedicated efforts carried out to adapt to the rapidly evolving technology. Patent protection and the monopoly of rights conferred through gives innovation a major impetus and hence, any regulation of AI will have to be done while being careful so as to not weaken IP protection available. This idea is at the centre of the response made by ACT to the UK IPO: legal approach towards regulation of AI should prioritise both providing technology neutral protection and also enabling AI to prevent and confront infringement of IP. IP protection is essential for a thriving digital economy and hence, the importance of a regulatory framework for AI should not make way for a weak regime of IP protection. ACT made the following important recommendations in its submission to the UK IPO:
IDENTIFYING THE INVENTOR
With respect to the identification of a the inventor of an AI invention, ACT’s approach is to divide the invention into three broad categories.- (1) a primary AI invention; (2) an alternative application of an AI invention; and (3) inventions developed solely by AI. The first category forms the baseline AI invention, the second is constituted of applications AI to additional contexts and the third category demonstrates unpredictable outcomes arrived at by AI itself. It is important to note that ACT considers AI primarily as a tool to be used by human inventors. We thus consider AI to be mostly a tool for human inventors to use. Now, baseline inventions are where AI is used to improve machine capabilities or function as an algorithm. These fit squarely into the mol of software inventions and their regulation in IP is not a problem. However, the picture is murkier when we foray into the realm of machine learning (ML). Here, a computer is trained to receive information and react based on previously fed commands, and there is no human intervention at the point of reaction. This is where AI begins to show signs of autonomy and the ability to produce original ideas. The final category of AI is an invention solely developed by a machine and has no human involvement, in such cases a human inventor cannot directly be identified.
RECOGNITION OF AI AS SOLE OR JOINT INVENTOR IN SOLELY AI INVENTIONS AND THE IMPACT OF ITS FAILURE ON FUTURE AI INVENTIONS
Even in cases where the invention is developed solely by the AI ACT proposes that the inventor would be the machine programmer whose intellect created the AI with the ability to resolve a human defined issue, an issue whose resolution required human scientific temperament. Furthermore, the lack of recognition afforded to AI as inventors would not have any bearing on the further development of innovation in the category of AI inventions. This flows from the fact that IP is rewarded to human inventors to act as an incentive to strive for more creations. There is a strong sense of self attached to IP, as also the satisfaction of monetary gains. These are distinctly human emotions and cannot be felt by AI. Therefore, it would not matter to AI whether or not it is being recognised as an inventor.
VESTING OF PATENT RIGHTS IN CASES WHERE AI RECOGNISED AS SOLE OR JOINT INVENTOR
This is an interesting aspect of the discussion If AI is recognised as an/the inventor, who or what would exercise the affiliated rights? ACT’s answer to this question is that this decision should be taken after examining the factual context of a given invention. Generally AI is seen as a tool used by human beings. Individuals who contribute to the conception, training, or applications of the AI may all receive consideration as AI inventors and in such cases specific rights would vest in specific individuals depending on circumstances surrounding the grant of the patent. These issues will have to be examined by the IPO during patent prosecution and only then an answer may be arrived at.
PATENTABILITY ISSUES IN THE FUTURE AS AI TECHNOLOGY DEVELOPS
The IPO should consider a variety of elements when evaluating and determining an AI invention’s patentability. Factors such as (1) data that is used to train the AI; (2) the algorithm; (3) the method of training the algorithm; and (4) the outputs produced thereof by the AI. The IPO should use the existing requirements for software patentability as a starting point to identify necessary elements of patentable AI inventions and applications.