Author: Francesco Ricotta

The first applications for invention patents designating artificial intelligence as the inventor have been recently filed with the European, US and British patent offices.

The first patent application claims a food container with hollows and protrusions characterized by fractal-shaped flush profiles delivering added stackability or joint resulting into ease of transportation and storage.

A few illustrations selected from the patent application for a food container showing fractal-shaped wall contours.

The second, instead, is about a warning lighting device consisting of a dimmable light source designed to flash according to a frequency based on characteristics fit to consistently locate it and suitable of being effectively intercepted by detection systems. The device thus stands out as particularly effective when the need arises to report an emergency.

While an international team of intellectual property professionals, researchers and consultants headed by Professor Ryan Abbott, who teaches at Surrey University, have been in charge of the applications filing, the findings underlying the applications seem to have been developed by an independent, man-free machine learning algorithm, called “DARBUS”.

According to the information so far shared by the initiative promoters and Dr. Stephen Thaler, creator of DARBUS (which, it is worth stressing, is in turn at the heart of an invention patent), the system consists of two distinct artificial neural networks mutually cooperating to process original concepts and to select the inventive scope, going as far as to designate the most promising ones deemed worthy of further development and finally leading to innovative solutions.

Providing a general information asset that ranges across various knowledge domains yet fulfilling no specific function, the first neural network is estimated to be able to combine the various notions underlying the knowledge asset in order to develop additional, differentiated concepts deemed (potentially) different and innovative exclusively on the basis of independent assessment.

The second neural network, defined “critical”, is estimated to monitor the activity of the first and to compare the concepts developed from time to time with the information asset known to the system in order to assess the different, innovative character and eventually give a positive or negative feedback to the first about the results achieved and, thus, influence later activities.

This having been said, it is sensible noting that, at the current state of things, the material provided by the inventor of DARBUS and the project coordinators does not clearly shed light on a few facets of the functioning of the algorithm and the process having led to the filing of the above mentioned patent applications.

Firstly, following the creative process, it fails to describe in what way the algorithm has consistently formulated the result of its inventive activity based on a format either suitable of being the object of a patent application (or just comprehensible to man). On the other hand, according to the details provided in the project website, reference is clearly made to the fact that patent applications regarding DARBUS-developed solutions have been drawn up by intellectual property consultants which further evidences that the contribution of knowledgeable professionals plays a crucial role when it comes to activities ranging from the development of an inventive solution all the way down to its coding according to a format suitable of being object of patent rights (irrespective of the inventor’s human or artificial nature).

In addition, a few critics maintain that the information concerning the system’s tangible functioning to date made available by the inventor of the algorithm and the team who have filed the above mentioned patent applications is not deemed enough to verify whether the system is gifted with an inventive ability of its own, also because DARBUS does not seem to have undergone the tests usually run to verify the functioning of artificial intelligence algorithms.

This consideration having been made, if the system was validated as capable of comprehending the state of the art currently known and independently processing new, inventive contributions, it would undoubtedly stand for a noticeable breakthrough.

Assuming that DARBUS artificial intelligence had succeeded in developing independently (without man’s intervention), technical solutions suitable of being protected as invention patents, it would be legitimate wondering the role such technologies play in the invention domain when we reckon that confining them to the mere role of “tools” in the inventor’s hands to provide a creative contribution would be questionable.

Indeed, the initiative seized by the team headed by Prof. Abbott seems aimed to unleash debate about the potentials offered by the use of such systems for the sake of creative and inventive processes as well as and, most of all, about the ownership of exclusive rights on creations resulting from the activity of such systems.

In this connection, it is worth noting that, from the judicial point of view, DARBUS’ designation as the inventor of the solutions underlying the above mentioned patent applications has limited implications since this designation would (theoretically) attribute artificial intelligence the right to moral paternity, namely being acknowledged as the inventor (doubting the exercise of which – at least at the current state of things – is legitimate), without any possibility to determine the economic exploitation. The ownership of the patent applications, in fact, has remained with a human subject (in the case in point, Dr. Thaler, DARBUS inventor and holder of the exclusive rights also in relation to the artificial intelligence system).

The initiative promoters claim that if, on one side, artificial intelligence should be acknowledged as the author of the inventive solutions developed independently, the ownership of the exclusive rights on the above mentioned solutions – including, first and foremost, those concerning economic exploitation – should anyway be limited to the owner of the artificial intelligence algorithm.

Such solution would foster the use of artificial intelligence systems to the aim of developing inventive solutions as it would allow to effectively protect the efforts and investments made by the authors of these systems, while reducing the risk that mere users of artificial intelligence may appropriate the inventive results achieved independently by these systems or with their very contribution. When we reckon with the numberless fields of application of artificial intelligence, this would stand for an overall innovation incentive.

Whether this thesis is endorsable or not, the ownership of intellectual property rights applying to artificial intelligence-driven creations or, anyway resulting from their very use, is going to be a much-debated topic in the near future, also in view of the ever growing number of applications of artificial intelligence systems in varying domains of knowledge.

Take for instance AI Portrait Ars designed by the Italian designer Mauro Martino. Based on a photograph, the system allows to create veritable portraits inspired to the creations of the great masters of the past yet completely different in terms of forms, colours and style. In this case, too, the system functioning is based on a neural network specially trained to carry out a given task which, according to the creator himself, would hamper any creative contribution from the network, thus raising no issues about the ownership of the works thus created.

Other than negligible, these aspects must be addressed in the framework of responsible, efficient business and intellectual property management, also in the light of the tremendous application potentials, the pervasive nature of these technologies and the current lack of established criteria to assess these aspects (which, as pointed out, are still being debated on an international scale).

Faced with such uncertainties, the mindful choice of a consulting company providing multi-disciplinary advice and on-going updates about highly topical issues like these is paramount. There is no denying that, even when it comes to inventions developed by tools like artificial intelligence systems, it is still crucial relying on proficient professionals effectively using the legal means available to claim rights about the inventive results thus achieved.

© BUGNION S.p.A.