The Need for Patenting AI Inventions
Rahul Dev, a Delhi-based patent attorney and technology business lawyer explains the need for patenting AI inventions, the current challenges and legal best practices to patent AI innovations.
The process of building an efficient IP strategy for Artificial Intelligence is also going up in recent years. Patent analytics company Parola Analytics that sees a 62% increase in patent applications related to AI, in the last decade alone, is proof of this. This rise could be attributed to the fact that in a world where AI-led inventions are on the rise, there is not only an intentional overlap of ideas but also a need to understand what the acceptable practices of patenting AI inventions are. However, several innovators still struggle to understand if AI can be patented and if yes, which set of rules or guidelines can be applied.
In an exclusive interaction with CXOToday, Rahul Dev, a Delhi-based patent attorney and technology business lawyer explains the need for patenting AI inventions, the current challenges and legal best practices to patent AI innovations.
CXOToday: With AI patenting assuming importance are traditional IP system and patents going out of fashion?
Rahul Dev: While several businesses continue to work around traditional IP systems and patents, of late, there have been discussions around AI patenting. Conventionally, tech companies developing hardware products were more inclined towards patent protection. However, with evolution of software solutions, IoT (Internet of Things) introduced the integration of hardware with software, which became a hot category of patents. Moving further, IoT resulted in generation of large amounts of data (big data) that required smart tools for processing and intelligence gathering. Such smart tools form the core of AI, and to ensure that AI keeps on improving on its own, ML came into existence.
AI patenting is essentially used to refer to two types of patents. First, patents covering innovations relating to AI products and services, such as, data analytics, self-driving cars etc. These innovations are developed to replace manual processes with automated ones for better productivity and efficiency.
Secondly, there is an exciting area of inventions developed by machines and robots instead of humans. Such inventions are known as AI inventions. The second type of patents brings to light a few questions. To give an example, if a robot creates a new product, can such product be patented by naming the robot as an inventor? This type of patenting has seen several developments on the issue raised wherein the European Patent Office (EPO) has denied two patent applications on the grounds that an AI system cannot be listed as the inventor.
CXOToday: Can an AI system be given a patent and how?
Rahul Dev: Yes, patents are regularly granted for innovations describing AI systems, subject to the criteria that such inventions are able to satisfy the local legal requirements pertaining to the patent eligibility.
Generally, laws across various countries (India, US, Europe) require three essential characteristics as the eligibility criteria to grant patent rights. These include: Novelty or newness of the innovation, Inventive step or non-obviousness and Industrial application. Among these, while almost all innovations qualify the industrial application requirement, the major challenge faced during patent examination is to convince the patent office about the novelty and inventive step.
Patent attorneys work closely with clients to navigate these objections. To counter novelty related objections, patent office needs legal and technical arguments to highlight how the innovation has additional features that have not been disclosed by existing patents.
However, the major challenge lies in convincing the patent office about the presence of inventive step as the patent examiners generally cite two or more references stating reading such references together makes the invention looks obvious. To handle these objections, patent attorneys submit to the examiner a set of arguments explaining how reading such references together is not the right way to decide the patent eligibility, specifically when there is no literature that mandates such combination of references.
CXOToday: Currently India is witnessing a surge in the number of start-ups, most of which are involved in the development of AI. However, there are no guidelines that can be applied to AI-related patent applications. What are your views on this?
Rahul Dev: The general guidelines for computer related inventions cover software inventions, mobile applications as well as AI related patent applications. For AI related patent applications, patent applicants need to highlight the core technical features that are unique as compared to competitors. Based on such features, patent attorneys can draft a strong patent application that can withstand the legal scrutiny at the later stage of patent examination.
To extend this further, there has been a recent Delhi High Court judgment to define the conditions based on which such patents can be granted. In the case of Ferid Allani vs UOI dated 12.12.2019, the Delhi High Court (DHC) held that patent applications in the field of computer programs would have to be examined in a manner to see if they result in any ‘technical contribution’. If the invention demonstrates a “technical effect‟ or a “technical contribution‟ it is patentable even though it is based on a digital platform, or a computer program. The DHC further stated that the effect which the computer programs (digital innovations) generate is crucial in determining patentability.
In addition, the court further stated that the term “technical effect‟ shall be interpreted according to latest practices of patent offices of foreign jurisdictions along with judicial precedents.
CXOToday: Machine learning gets maximum focus in AI patenting. What are the other areas and scope?
Other than machine learning, we feel data processing and analytics are also extremely focused upon. These areas are also focussed upon deeply as eventually AI and Machine Learning involves processing and analysis of data. This is evident because for any AI tool to evolve and perform better with time, the AI tool needs to learn from the data provided so as to reduce the chances of errors. Such learning occurs due to introduction of ML tools, which improve the performance of AI tools.
CXOToday: What are the challenges around patenting of AI innovations?
Rahul Dev: One of the largest issues AI faces is the inability to be defined or understood. A huge debate remains whether AI should be patented, or whether it should be protected merely as a trade secret of the entity behind its creation. This leads to the challenge of AI ownership. A lot of this is the work of genius, intensive education, years of research, and dedicated hard work. Hence, the obvious question asked is – who owns this research, and who gets to benefit from the credit of it?
Also, over the years, we have noticed that companies working on AI innovations usually focus on automation of business processes. For many companies, the goal is to reduce the dependency on the human resources and develop AI tools to achieve the desired goals in the form of machine learning.
From the perspective of patents, legal protection can be availed for both the AI, as well as the ML features. The Indian Patents Act specifies that patent can be granted for an invention capable of novelty (newness), inventive step (non-obviousness), and industrial application.
Companies face challenges at the very first step of determining whether their innovative products and services qualify for patent protection or not. Hence, the best strategy is to engage patent experts to develop best internal practices for invention management as a proactive approach.
CXOToday: How do you empower CIOs and tech innovators to realize the benefit of patenting AI?
Rahul Dev:The CIOs and tech innovators can extract the benefits from patenting AI by having strong internal processes. These processes can enable innovative ideas with an organization to reach the management from the creators, which can then be filtered to decide whether it makes sense to file AI patents.
Considering the overall business landscape after Covid19, the tech industry is all set for a revamp. Remote working has become a norm and future M&A activities will focus significantly on tools improving the remote work. Such tools will definitely have AI and ML features, which if patented, can add immense value to the business as these include the intangible assets for any company.
CXOToday: What could be the legal best practices to patent AI innovations?
Rahul Dev: The best practices should include keeping a close eye on the latest court decisions and patent office orders or guidelines that recognize that digital innovations, including AI and ML innovations may not necessarily include invention of physical hardware elements. Therefore, the AI and ML features are to be presented to the patent office in a manner such that these are able to satisfy the legal requirement.
CXOToday: Please throw some light on the future of AI patenting.
Rahul Dev: As seen from the latest developments at the US and EU patent offices, the next wave of legal issues surrounding the patent process will be on the inventive features or products created by AI instead of humans. There will be a pressing need for the appropriate laws and regulations to define whether non-human inventors can file for patents or not.