Can write AI patents

Institute for Applied AI

Patents in the development of AI technologies - stimulation or blocking?

(Shutterstock / Gorodenkoff)

Online event, Tuesday, January 12, 2021 - 5:00 p.m.

You can download the slides of the presentation here.

International patent disputes on "Connected Car" technologies as well as the broad discussion on qualifying AI as "inventors" provide exemplary evidence that questions of intellectual property have also affected the dynamically developing area of ​​AI. In law and innovation research, patent protection finds its legitimacy in particular in the reward character for the promotion of technological development through the disclosure of an invention. Patent protection allows the disclosure of technological developments without loss of protection for the inventor and thus enables scientific discourse and further developments.

At the same time, by securing the rights to new developments, patent protection enables cooperation and licensing models to bundle development efforts. This applies in particular to cross-sectional AI technologies, in which AI methods are used for various applications. However, these advantages of patent protection, which stimulate development progress in the AI ​​field, are also offset by risks. For example, patent protection creates prohibitive rights and the actors in the AI ​​sector are confronted with possible patent infringement claims and the resulting liability risks. A look at patenting practice shows that an enormously increasing number of patent applications can be observed both with regard to AI techniques (machine learning etc.) and with regard to functional means (language processing, visual detection, etc.) and AI applications. According to a study by the World Intellectual Property Organization (WIPO) from 2019, an annual growth rate of 175% can be determined for the area of ​​machine learning technologies alone, which is well above the average growth rate for patent applications in other technology areas.

In the international race for AI patent applications, the US patent office is still just ahead of the Chinese patent office, but there is a sharp increase in AI-related patent applications in China, which is often limited to China. It is not uncommon for such AI patents to be used to establish license models, but a steadily increasing number of active patent enforcements aimed at monopolizing AI applications can also be determined. Since no solid decision-making practice for patenting AI technologies and for determining the scope of patent protection has yet been established, legal uncertainties arise that are only slowly resolved by the patent offices and courts. Unjustly granted patents or an unjustified allegation of infringement can also have a behavioral effect on AI developers or AI users, since the attack, the ability to protect or defend against infringement claims is associated with prognosis and cost risks, which not only prevent startups from promising developments.

Determining a benchmark for granting patent protection for AI technologies has to balance these advantages and disadvantages. The intensive discussion on software patents and open source models in the past provides points of contact that can be made usable, but there are also specific features in the AI ​​area that exclude the simple projection of principles established in the past. The patent offices in Europe, the USA, Japan, China and other countries are currently trying to establish standards for patent protection for AI technology through guidelines for granting practice. However, such efforts will not prevent that there will be a large number of patent disputes in this economically extremely relevant market in the future, which will affect actors in science as well as in business. The task in the next few years is to maintain the stimulating function of patent law on the one hand and to prevent innovation hurdles caused by patent risks on the other.

A lecture at the IAAI on this topic will deal in detail with patenting practice and identify possible practical areas of conflict both in commercial exploitation and in scientific research. Furthermore, the lecture will deal with possible cooperation models in the field of AI development. The lecture should also provide an impetus for a new research project at the IAAI.

Online event, Tuesday, January 12, 2021 - 5:00 p.m.

Live Stream: To receive the conference link, please register here.