Author: Preslav Baldzhiev
With the development of digital technologies, humanity is increasingly turning its attention to Artificial Intelligence (AI). Currently, AI is used in various fields – healthcare, it is built into more and more devices for everyday use (mobile devices, computers, household appliances), and it is becoming a major component in the automotive industry. Despite its wide scope, AI remains an area that is poorly regulated, and this threatens the legal regime that governs it.
What is AI?
AI is the result of concepts that allow computers to take human-reasonable actions. These concepts are based on the idea that human intelligence can be described to such an extent that it can be reproduced by a machine. In other words, AI is a set of algorithms, mechanisms, and systems that allow the computer to self-learn. Such machines are defined as “conscious”, and they are called intelligent agents.
Аt the moment, there is no single objective criterion for determining which machine has the quality of “Consciousness” – for now, the most used is the Turing test, which relies on subjective human judgment.
AI as an object of Intellectual property
The European Patent Office accepts AI as a set of computational models and algorithms within the meaning of the European Patent Convention (EPC). Such mathematical algorithms cannot be patented unless they represent a technical effect that is relevant to the whole system of the invention. But even if they represent such technical effect, they will again not be patented on their own, but together with the system in which they are integrated (they will be considered in their entirety).
The technical effect is related to the functions of the device, through which the goals/results for which the device is created are achieved. In the context of this article, the technical effect will be expressed in the self-learning of the computer system, which is possible due to the mathematical algorithms embedded in it (the computer system and the algorithms will be considered as one).
Аn example of patentable objects with AI is the models of the car brand “Tesla”, which have the so-called “autopilot” – a set of algorithms that allow the car to move freely on the roads without human intervention. The algorithm itself will not be subject to a patent, but the algorithm and the car (as a whole) will be.
AI as a subject of Intellectual property
There are examples in which AI systems become patent applicants. The most striking example is the AI DABUS, which is listed as an applicant for two patent applications addressed to the Patent offices of the United States and the United Kingdom, as well as to the European Patent Office (EPO). The three patent offices come up with the same decision – AI cannot be a patent holder, as DABUS does not meet the legal requirements for an inventor.
Currently, the three patent offices accept that an inventor can only be a natural person. An additional obstacle to the successful registration of AI as an inventor is the lack of a document to authorize the real applicant – the applications in which DABUS is named as an inventor are submitted by the creator and owner of the AI.
The EPO’s decision has also been supported by its Legal Board of Appeal (case J8/20). The decision, issued by the Board of Appeal, is based on the EPC, according to which inventor can only by a natural person.
This is a possibility that is not provided by the law – with the filling of this void, however, such a waiver of a patent for an invention created by AI should not be ruled. These considerations can be supported by the following example.
Contrary to the earlier decisions, the Australian authorities concluded that artificial intelligence could and had been successfully registered as a patent holder. The example is again related to DABUS – the Australian patent authority rejected the request to register Dabus as a patent holder, arguing that an inventor could only be a natural person (this statement is supported by the definition of the word “inventor” in the official local dictionary). This decision is subject to judicial appeal, and the decision of the competent Australian court is just the opposite – in the reasoning, the judge argues that the definition of inventor in the official dictionary is outdated and presents an archaic understanding that does not meet technological and information reality. Another argument is that Australian law does not explicitly prohibit artificial intelligence from being an inventor and patent holder.
Such a decision can be called a “scientific breakthrough”, as it is the first, but it is unlikely to be the last to allow artificial intelligence to be the holder of intellectual property rights. However, the issue is still controversial due to the lack of unanimity among the competent authorities.
As for the copyright of works that are created by self-learning machines, there is no consensus. This question is currently extremely theoretical due to the ambiguity of whether artificial intelligence can exhibit “creativity” and whether the work created by AI would have the quality of “original”.
Conclusion
The improvement of AI is far more dynamic than the development of the legislation that governs it. This creates a vacuum that seriously hinders the work not only of enthusiasts dealing with this technical challenge, but also hinders the stability of civil turnover.
This material prepared by Preslav Baldzhiev aims to provide more information about artificial intelligence and intellectual property. It does not constitute a legal opinion and cannot be interpreted as individual consultation on any concrete facts or circumstances. The advice of a specialist should be obtained for specific questions and situations. For more information on the above-mentioned issues and individual consultations, please contact the team of the law firm of Krasimira Kadieva at 00359 882 308 670 or make an inquiry using the contact form of the website. Since 2017 Preslav Baldzhiev is a law student at Sofia University “St. Kliment Ohridski “, having previously graduated from the High School of mathematics and natural science “Acad. Nicola Obreshkov” in Burgas. In February 2020 he took a course for industrial property representatives at the Patent Office of the Republic of Bulgaria in the field of trademarks, geographical indications, and industrial designs. He is interested in intellectual property, personal data protection, commercial and law on obligations and contracts and also regularly attends conferences, practical courses, seminars, and webinars.