India braces for the merger of artificial intelligence and IP
30 April 2020
India must act now to sync its intellectual property system up with the growth of artificial intelligence, a lawyer says.
A December 2019 report titled Understanding the Dynamics of Artificial Intelligence in Intellectual Propertypublished by Tata Consultancy Services, India’s largest software exporter, predicts that in spite of the advancement of patent laws in India, the growth of artificial intelligence across the world entails the need for new policies for the enforcement of intellectual property rights.
“Current patent laws treat AI software inventions essentially as logical algorithms implemented on the computer. While patent eligibility of algorithms is valid, there is little about how to deal with inventions that are heuristic in nature,” the report says.
Mohan Dewan, principal at RK Dewan & Company in Mumbai, agrees, saying that India’s IP laws are not in sync with the developments in AI.
“AI software is treated as a logical algorithm implemented on a computer,” he says. “While algorithms can be patented, that is not where AI ends. AI inventions are heuristic in nature, which means that it uses a technique which is faster than classical methods. Previously, common types of software were rule-based systems, but they have become heuristic, displaying higher degrees of intelligence over rule-based systems. There is nothing in patent laws as to how to deal with inventions that are heuristic in nature. In AI, machines derive solutions either on their own or in association with a natural person, which brings into question the definition of a natural person. This area remains unaddressed.”
Dewan says that that in an integrated world, data may move across geographical locations and may be handled by various agencies, involving issues of data privacy and ownership. Then there are issues related to who shall own IP rights on inventions – the data owner or the AI scientist.
Among the issues India faces when it comes to artificial intelligence in intellectual property, according to Dewan:
Inventor and ownership issues. IP recognizes patents as a creation of the mind. “We all know that many AI inventions are the creation of a machine,” Dewan says. “In fact, an AI machine can learn from experience, draw its own inferences, identify differences and even develop its own language. So, an AI can be a complete creation of the machine. But patent laws the world over recognize IP as a creation of the mind, and a machine cannot be considered to have a mind of its own. Therefore, it raises the question of who owns the patent in case it is created by a machine.”
Person skilled in the art. Patent laws of most countries suggest what a person skilled in the art would derive from a particular invention or what conclusions a person skilled in the art would derive from prior art. A person skilled in the art is a person who does not have the level of knowledge of an inventor but has ordinary knowledge in the subject matter of the patent. “If the major contributor in case of AI invention is a machine, whom do we say is a person skilled in the art and how would one evaluate novelty and inventive step?” Dewan asks.
Subject matter eligibility. Most countries have excluded abstract ideas, laws of nature and natural phenomena from patent-eligible subject matter, and many an AI subject matter may fall under such exclusions. For example, in the medical world, AI has been found to be extremely successful in diagnosis and prognosis, relating images with known images, drawing extrapolations and conclusions based on experience, just as would be done by a practitioner in the field, of course at much lower efficiency. Similar situation could exist in many other domains. So, the activity is directed to a mental process and the abstract concept of using mathematical algorithms to perform certain functions, which may not be patentable. AI can draw comparisons with huge sets of data and draw conclusions or modify them with experience, which humans have been doing for years together and therefore, lack inventive concepts.
Since India is advantageously placed when it comes to AI patents and AI skills, the governmenthas already prioritized AI as a thrust area. Most developed and developing nations are investing heavily in AI and India needs to do the same, says Dewan, who believes that while there are individual areas of brilliance in AI in India, collaborations between industry and academia are missing.
According to Dewan, the government’s NITI Aayog policy think tank has recommended setting up the following centres for taking AI to the next level:
- International centre for new knowledge;
- Centre for research on sub-systems;
- Centre for advanced studies, translational research and leadership, focusing on development and deployment of application-based research; and
- Centre for excellence in technology innovation and transfer.
“It also plans to equip existing manpower with necessary skill sets and to introduce curriculum in colleges in line with AI requirement skills,” says Dewan.
Dewan says that what is expected of AI in IP in the near future is that basic information about the new invention may be fed into a system, which shall do a thorough search and identify whether the invention is novel and has an inventive step.
“Subsequently, if that be so, the system shall be able to draft complete specifications in terms of guidelines fed into the system, including the claims,” he says. “So, the first draft of a patent application may be ready in a jiffy and a human may simply have to review the draft. The system can also generate probable questions that are likely to come during examination and what are the possible corresponding answers. Such applications may have challenges to overcome initially but they could be overcome with passage of time. It shall also be possible for an AI system to have the extracts of all court cases in the backdrop and take decisions about drafting accordingly. The AI system could set reminders for various deadlines, fill forms automatically and send them to an appropriate agency. An AI system could use various IP algorithms, NLP algorithms, semantic technologies, automated reasoning, text mining, among others. in order to achieve what has been discussed above. In other words, it is likely that IP property management may be possible with minimal human intervention with the help of AI.”
Excel V. Dyquiangco
