Robot inventors are on the rise. But are they welcomed by the patent system?

Artificial Intelligence is now capable of performing tasks that have historically required human ingenuity, such as composing musicals, conducting orchestras, generating innovative ideas and drafting patent applications. This progress has the potential to improve efficiency and make savings across all industries. But as detailed in a new white paper by the World Economic Forum's Center for the Fourth Industrial Revolution, these advances also put AI on a collision course with numerous aspects of patent law.

Once praised by Abraham Lincoln as one of the three great advances in world history, US patent law has thus far been able to achieve its main objective of enhancing social welfare through the promotion of innovation, dissemination of useful technical information and investment in new and beneficial technologies. But the anticipated disruption of AI will be far more pervasive than previous technological changes and will challenge the core legal standards that serve as the safeguards of patent law. Insufficient preparation for this could result in an outdated patent system that is no longer able to fulfill its intended functions.

The intellectual property system was established with “the conviction that it is the best way to advance public welfare through the talents of authors and inventors in ‘Science and useful Arts’” as mentioned in the Supreme Court decision in Mazer v. Stein in 1954, and is the only policy mechanism expressly designed by the US Constitution to promote innovation. A failing patent system can have significant and adverse effects on innovation and social welfare.

For example, if the patent system allows AI-generated inventions to be patented without any meaningful human oversight, could that result in too much power in the hands of a few leading companies that own those AI systems? We therefore need to find ways to help the patent system evolve so it can continue to perform its main objectives. These dialogues must involve patent law’s relevant actors - legislators, judges, academics, and practitioners - and other stakeholders, such as scientists, entrepreneurs, investors and technologists, as well as others with meaningful expertise in innovation and ethics.

The crucial patent law issues that face impending disruptions and merit discussion are, firstly, whether the present standard on patent subject-matter eligibility requires any changes to promote AI innovation; secondly, whether AI-created inventions should be patentable; and finally, how liability should be assessed when AI infringes on a patent claim.

Patent law does not provide protection for inventions directed to certain areas. In the US, for example, abstract ideas, natural phenomena or laws of nature cannot be patented, because they are considered as “basic tools of scientific and technological work” and monopolies over them would impede innovation. And the recent Alice decision has made it more challenging to obtain patents on software and “computer-implemented inventions” which often get characterized as abstract ideas. This has also made it more difficult to obtain patents on a subject matter that can be performed through an “ordinary mental process,” “in the human mind” or by “a human using a pen and paper.”

This trend has created tension for patenting AI technologies because their goal is often to replicate human activity. Given their innovative potential, there must be further discussions on whether the presently-heightened standard promotes or stifles innovation. A good starting point for these discussions are the present debates on the subject-matter patentability standard with regard to software, but these must then evolve to account for AI-specific considerations.

If the present standard is found to have a materially negative impact on AI technologies, then discussions must focus on making adjustments to the standard so it can better achieve patent law’s main objectives.

The discussions should also focus on balancing the promotion of innovation with furthering societal benefits wherever possible. One option, for example, would be to lower the subject matter standard only for AI inventions that relate to areas such as healthcare, the environment, criminal justice and education.

The question of whether inventions generated by AI are patent-protectable has not yet been expressly answered, although US laws generally frame patentability and inventorship in terms of human creation.

Tracing the origins of US patent law, Thomas Jefferson stated that an “inventor ought to be allowed a right to the benefit of his invention for some certain time” to encourage “men to pursue ideas which may produce utility.” Consistently, under US patent law, an invention requires conception, which is “the formation in the mind of the inventor, of a definite and permanent idea of the complete and operative invention,” where the “inventor” refers to an “individual.” The Federal Circuit explained that to “perform this mental act, inventors must be natural persons and cannot be corporations or sovereigns.” The remainder of the Patent Act is also replete with references to human actions. When the Patent Act was put in place, there was likely no need to characterize the inventive process as being performed by anything other than people, because there were no such other “beings”.

But that is no longer the case: Stephen Thaler’s Creativity Machine generated an invention that was later issued as a US patent. Likewise, John Koza’s Invention Machine has been acknowledged for having produced an idea that culminated in a US patent. In both cases, however, the involvement of AI technologies was not disclosed to the US Patent & Trademark Office and only humans were listed as inventors.

Granting patent rights to AI-created inventions may expedite innovation, even enabling advances that would not have been possible through human ingenuity alone. But it is equally vital to acknowledge the potential negative effects, such as an atrophy of human intelligence.

Possible middle grounds between these competing interests must be identified to help the patent system achieve its main objectives in a well-balanced manner. For example, if AI-generated inventions are ultimately deemed patentable, would granting shorter patent terms to them help to even the playing field between AI and human inventors?

AI has the technological capacity to infringe patent claims so we must discuss how to address liability in such cases. Who should be held responsible and how should damages be assessed? Should humans, such as the AI’s user, developer or proprietary owner, be held responsible? Would this be fair if AI becomes truly autonomous?

This European Parliament resolution can serve as a helpful guide on the discussion of liability. Of particular note is its discussion of implementing an obligatory insurance scheme or recognizing AI as a legal person for civil liability and other purposes.

The different, existing liability frameworks, such as strict liability, negligence models or products’ liability, must be analysed for their relative strengths, while new approaches are researched to see if they can function more effectively.

The white paper also explores other patent law topics that are likely to be disrupted by AI, such as AI’s role in the definition of “a person of ordinary skill in the art”, the need for regulating patent applications prepared by AI, the treatment of AI-generated content as prior art and the Patent Office’s use of AI as a helpful tool. These emerging legal issues pose challenging questions, but it is essential that they are proactively researched and discussed by the relevant actors and stakeholders, so effective solutions can be provided to promote innovation in a socially inclusive and ethically-responsible manner.