After a few cases made headlines recently, IP expert Richard Johnson asks whether artificial intelligence can be recognized as an inventor under current patent law.
The increased use of AI-based systems as tools for promoting innovation raises fundamental questions about the concept of inventiveness. Court rulings in a number of countries provide an up-to-date insight into the understanding of the term âinventorâ in patent law.
What is the problem?
The most recent legal proceedings all stem from the activities of the Artificial inventor projectthat supports tracking of two patent applications for concepts created with a certain type of AI system called DABUS.
The DABUS system uses a kind of reinforcement learning technique in which two AI subsystems (so-called artificial neural networks) interact with each other.
The first is created with a âworldviewâ made up of data contained in one or more information sources. It is trained to learn links between different topics and is then configured to independently disrupt the links made during the training.
The second monitors the first in order to recognize the disturbances as new ideas and to evaluate them for their usefulness or some other value in relation to the already existing worldview. It then brings those who have a benefit back into the world where they can be evolved through further disturbances. Ideas without a benefit are not fed back, so the second AI subsystem works in such a way that the new ideas with a perceived benefit are promoted.
The patent applications force a discussion about the inventorship by identifying the DABUS system itself as the sole inventor.
The DABUS patent applications were rejected by the patent granting authorities in a number of countries for failure to comply with formalities for identifying an inventor. These refusals have been challenged in court and the results currently fall into two camps.
The law states that an inventor is a person
The European Patent Office (EPO), UK and US courts believe that current patent law only allows one person to be named as an inventor. The decisions share an essential reason why the wording of the law clearly indicates that the term âinventorâ refers to a person.
The justification of the EPO is based in particular on the idea that the legal claim to own a patent actually originates from the inventor. The judicial rulings in the UK and USA were based primarily on an analysis of the wording and history of the relevant legislation.
The decisions of the EPO and the UK did not deal directly with the more fundamental question of whether or not a machine could actually invent. In contrast, the U.S. court ruling referred to a report by the United States Patent and Trademark Office (USPTO) following extensive consultation on the impact of AI on the intellectual property system (shortcut). One of the general conclusions in this report was that AI is not mature enough to support a change in patent law for political reasons.
A contrasting look
Contrary to the rulings in the US, the EPO and the UK, an Australian court was the first to rule that an inventor (for the purposes of Australian patent law) does not have to be human.
In the Australian decision, the judge took the view that it was appropriate to change the interpretation of the relevant provisions as technology evolved. He justified his interpretation in part with a “statement of object” in Australian patent law, which reinforces the idea that patent law is essentially there to disseminate information and promote economic well-being. The judge was of the opinion that protecting AI-generated inventions would serve this purpose and therefore supported the decision.
The decision has met with some criticism, especially as it risks establishing Australia as an outlier against the USPTO and EPA, which are among the most influential patent granting agencies.
What is in store for the future of AI inventiveness?
The decisions made so far are primarily about the interpretation of legal regulations and not about the fundamental question of whether a machine can invent or not.
Currently, the direction of travel is such that applicable law is not flexible enough to accommodate non-human inventors. This should bring the debate back to the policymakers, and this debate should address the issue of actual AI creation.
In reality, the involvement of an AI system in a particular invention is likely to range between completely “non-inventive”, for example, by requiring some form of human interaction, and completely “inventive”, if no human interaction is required at all. This spectrum is not well defined and the court rulings do not provide any guidance on how inventorship is determined. The future of the âinventive AIâ debate seems to lie in this area, as courts are generally reluctant to adapt the interpretation of current legislation.
Richard Johnson is a patent attorney with an IP firm Mewburn Ellis