Why ‘No Patent’ for AI-made Inventions is the Biggest Blunder?

Intellectual property regulations, as well as an international convention, are required to patent machine-created inventions

A machine-learning algorithm assisted researchers in developing a powerful antibiotic that works against a wide range of diseases in 2020. AI is also being utilized to help with vaccine research, medicine design, material discovery, space technologies, and ship design. AI might be used in a variety of technologies during the next several years. This is posing one of the most serious threats to patent systems in history. Patent law is predicated on the notion that inventors are human; it currently struggles to deal with a machine inventor. Courts throughout the world are currently grappling with this issue since patent applications citing an AI system as the inventor have been filed in over 100 countries. Several organizations are holding public discussions on artificial intelligence and intellectual property (IP).

If courts and governments rule that AI-created ideas cannot be copyrighted, the consequences might be massive. When the return on investment is restricted, funders and corporations are less likely to undertake beneficial research utilizing AI innovators. Society may miss out on the creation of valuable and perhaps life-saving inventions. Rather than pushing traditional patent rules to adapt to new technologies, we suggest that national governments create customized intellectual property legislation — AI-IP — to safeguard AI-generated ideas. Nations should also draft an international convention to guarantee that these laws adhere to uniform principles and that any conflicts are quickly handled. Both processes must be informed by researchers.

Who, not what — matters

Machines with the ability to innovate were not taken into account by the authors of the world's first patent legislation, the Venetian Patent Statute of 1474. They were also not considered in the 1883 Paris Convention for the Protection of Industrial Property, which laid the groundwork for the worldwide patent system. Even in 1994, when the World Trade Organization concluded its Agreement on Trade-Related Aspects of Intellectual Property Rights(TRIPS), AI-generated ideas were nearly unheard of. Today, worldwide patent rules are mandated by the treaties of 1883 and 1994.

TRIPS protects "all innovations, whether goods or processes, in all disciplines of technology, provided, they are novel, include an innovative step, and are capable of industrial application." 'Inventions,' 'novel,' 'inventive step,' and 'capable of industrial application' are all words of have legal definitions. In essence, if any of these conditions is not satisfied, an item will not be patented. The machine utilizing artificial intelligence cannot be labelled as an inventor because, under US law, only a person may be an inventor, a federal court said in the first American decision in a global controversy over how to designate computer-created creation. According to U.S. District Judge Leonie Brinkema in Alexandria, Virginia, federal law requires a "person" to swear that he or she is the inventor of a patent application, and both the dictionary and legal definitions of a person are natural persons.

What exactly is patentable?

Before an invention may be patented, it must generally fulfill all of the following standards.

• A creation by one or more inventors – This encompasses practically all technological items, processes, and procedures.
• Unique – The innovation does not exist yet.
• An inventive or non-obvious step – A 'person competent in the art' with 'common broad knowledge' in that subject would not recognize the innovation.
• Capable of industrial or utility use – The innovation may be manufactured or employed in industry, performs the claimed function, and/or has economic importance.

These principles, established by the 1994 Agreement on Trade-Related Aspects of Intellectual Property Rights, must be followed by all 164 World Trade Organization members (TRIPS).

New technologies have previously tested the system. High-profile cases have examined whether genetic sequences, living beings created by humans, and other items can be patented. In these cases, the primary legal challenge was whether they were innovations at all. For example, following a years-long legal battle in Salt Lake City, Utah, between the US Association for Molecular Pathology (among others) and molecular-diagnostics firm Myriad Genetics, the US Supreme Court ruled in 2013 that isolated human gene sequences are unpatentable because genetic information is a product of nature rather than a human invention.

AI-generated discoveries provide a new challenge to the patent system since the issue is determining "who" originated rather than "what" was produced. The first and most important question that patent offices have had to address in light of such changes is whether the author must be human. If this does not happen, one concern is that AIs will become so numerous that the patent system would be overwhelmed.

Another problem is of a more basic kind. An 'inventive step' happens when a 'person skilled in the art' deems an invention to be 'non-obvious.' This hypothetical person has average expert-level aptitude and understanding of the relevant technical discipline. If a patent examiner determines that the invention is beneficial, it will be granted. If a patent examiner determines that the invention would not be evident to this hypothetical individual, the innovation will be one step closer to being patented.

However, if AIs become more informed and skillful than everyone in a profession, it's unclear how a human patent examiner will determine if an AI's idea was apparent. Before the invention, an AI system developed to evaluate all published material about a field of technology would have access to a significantly bigger corpus of knowledge than any person. Almost everything would appear clear if measured against all knowledge. If everyone in the future has access to AI tools, the 'inventive step' requirement for patentability will be nearly hard to meet, and essentially nothing will be patentable. It would necessitate a full rethink.

Test case

Patent registration agencies in the United Kingdom, the United States, Europe, South Korea, Taiwan, New Zealand, and Australia have all rejected applications thus far. The majority of legal challenges to these choices have failed, with courts ruling that inventors are supposed to be human. In Germany, a court agreed that the ideas may be protected if Thaler was credited as the inventor who pushed DABUS to make them – a compromise that recognized the AI system's involvement. However, the tide of court opinion is virtually wholly against AI systems being recognized as inventors for patent purposes at the current time. Patent registries and courts are now forced to interpret and apply existing legislation as best they can in the lack of explicit regulations defining how to evaluate AI-generated innovations. This isn't great in any way. Governments should design laws that are specifically adapted to AI creativity.