The United States Patent and Trademark Office (USPTO) recently denied permission to list an Artificial Intelligence (AI) system, e.g., a robot, as an inventor. The 22 April decision came in response to a petition filed by FlashPoint IP Ltd regarding an invention that was created by an AI system called DABUS. In rejecting the application, the USPTO explained that under current law, only natural persons may be named as an inventor in a patent application.
The USPTO cited title 35 of the United States Code as consistently referring to inventors as “natural persons” and so an AI system cannot be named as an inventor. The United Kingdom’s Intellectual Property Office (IPO) and the European Patent Office (EPO) have also ruled that an AI system cannot be listed as an inventor based on similar legal interpretations. In light of these decisions, there is an intense ongoing debate: Can a robot be an inventor?
The USPTO issued a request for comments on Patenting Artificial Intelligence Inventions, on August 27, 2019. One of the questions posed is: “Do current patent laws and regulations regarding inventorship need to be revised to take into account inventions where an entity or entities other than a natural person contributed to the conception of an invention?’’
While the FlashPoint decision suggests that the USPTO has not changed its position on allowing only natural persons as inventors, many new innovations are aided or created by artificial intelligence, and many unanswered questions remain. One among many is can an AI system or robot truly invent? The USPTO Manual of Patent Examining Procedure (MPEP) states that “an invention is made when there is a conception and a reduction to practice”.
The MPEP defines conception as “the complete performance of the mental part of the inventive act” and it is “the formation in the mind of the inventor of a definite and permanent idea of the complete and operative invention as it is thereafter to be applied in practice.” It is further stated that “conception must be done in the mind of the inventor”.
Second, is an AI system truly capable of conception, as stated in the MPEP? While AI systems solve many problems that they are trained to do based on the data that has been fed into the system, it is unclear if an AI system can have independent thought. An AI system does not exhibit “common sense” unless that common sense has been built into the system.
AI systems are trained by feeding large amounts of cause and effect data. Multiple experiments at universities like Massachusetts Institute of Technology (MIT) are being conducted to codify the basic concepts and rules on how the world works. The data from these experiments are fed to AI systems to enable them to perform human-like reasoning when confronted with novel situations. However, despite years of research, the AI systems are a long way from achieving this objective. In such a situation can an AI system conceive and create an invention, without being programmed by a human on the specific nature of the problem?
Does it have independent thought and innovative capability other than what is programmed? Would the AI system understand the difference between causation and correlation? These are questions that need further debate. Also, the question arises if a robot have a unique identity to have a right to inventorship? Assuming that a robot is capable of conceptualising and thereby inventing, the question of whether a robot has an independent identity arises.
Would two robots manufactured identically have unique identities and unique behaviour? If not, how would one differentiate between the two? One could argue that an AI system does not have free will, and thereby not capable of exhibiting behaviour for which it has not been programmed or trained. In the absence of free will, one could argue that the AI system does not have a unique identity.
In 2018, the United States Court of Appeals decided that a monkey cannot own copyright to his selfie. One of the concurring opinions in the decision states, “There is no grounds to conclude ‘person’ in the US statute that could include anything other than natural persons.” There’re many ethical concerns too while taking robots are inventors.
Robots are not employees that can assign their rights in an invention to an employer under an employment agreement. Human employees offer their skills to their employer, i.e., work-for-hire, for an ongoing monetary consideration and any invention related to the person’s work-for-hire is generally owned by the employer. But employees are not “property” of the employers who does not own every idea conceived by the employees.
However, robots are the property of their owners and do not have rights of self-determination. In effect, if the robot is given right to inventorship, it would be hollow as the inventorship right would be without the rights attributed to a free human being. And this is also without legal and procedural challenges? An inventor applying for a patent has to sign a declaration under oath that the person is the true inventor.
It is not possible for a robot to sign such a declaration. At best, the owner of the robot could do so, but the owner cannot swear on behalf of the robot as the owner has not conceived the idea behind the invention. Under the US patent law, an inventor is the owner of the invention until the inventor assigns the ownership rights to someone.
A robot has no ownership rights in an invention. The fundamental concept of the US patent law that an inventor is an owner of the invention until the ownership rights are assigned to some someone simply does not work if a robot is an inventor. A human inventor assigns the ownership rights to someone for a monetary consideration. This consideration could be salary.
A robot cannot receive monetary consideration for assigning its ownership rights in the invention to the owner of the robot. Therefore, the question of whether a robot has assigned its ownership rights to the owner of the robot is moot. In addition, during patent litigation, there could be a need for the human inventor to be questioned in a deposition based on the past recollections of the inventor.
If a robot is granted inventorship, then it would not be possible to depose the robot based because robots do not have “past recollections” like human do. In summary, giving AI systems or robots inventorship rights is fraught with difficulties. There are many issues that even changes in the rules and legislation might not be able to address. Time for countries like India, now in an urgent state to use AI for many crisis-control situations as in the case of Covid-19 currently, to learn from the experiences of countries like the US and the UK to chart the tech-journey ahead.
Raj Dave is an IITian-turnedpatent attorney based in the US and Krishna Shastri is an IITianturned-patent consultant based in Bengaluru.