The technological revolution leading the current historical period has ushered in a new era, that of Artificial Intelligence.
Studies on the subject have progressed to the point where it is now possible to find androids capable of performing activities that so far only humans ever performed.
If we consider, in particular, that there are highly sophisticated machines capable of producing paintings or pieces of music objectively indistinguishable from human productions, it is clear that the impact of Artificial Intelligence (AI) on Intellectual Property law and, in particular, on Copyright is considerable and destined to be increasingly impactful.
Artificial Intelligence – Deep Learning
John McCarthy, one of the founding fathers of artificial intelligence, defined the discipline in 1955 as the process “of making a machine behave in ways that would be called intelligent if a human were so behaving”.
The World Intellectual Property Organisation (WIPO) defines AI in the context of Intellectual Property as a “discipline of computer science that is aimed at developing machines and systems that can carry out tasks considered to require human intelligence”, with little or no human intervention.
At the heart of an AI system is its reasoning/information processing algorithm, which takes data from sensors as input and proposes an action to be taken, given the goal to be achieved.
Among the techniques that most attract the attention of the IP lawyer are those based on machine learning, including deep learning.
Deep learning, in particular, is the form of machine learning that allows computer and digital systems to learn information more in depth, just as it happens in human learning processes, with the aim of more efficiently reducing the margin of error in the tasks that humans perform.
Consider the four-armed robot “Shimon”, created by Gil Weinberg, director of Georgia Tech’s Center for Music Technology in Atlanta. Shimon, after listening to several musical genres including pop, rock, classical and jazz and over 5,000 songs from Beethoven to the Beatles to Lady Gaga, is capable of writing and composing music by playing the marimba.
Not to be outdone is Xiaoice, a chat bot developed by Microsoft that has produced the first collection of poems written by a machine with artificial intelligence
Or again, Busker, a robot-painter created in 2015 by Paolo Gallina, professor of applied mechanics and robotics at the University of Trieste, which is able to create works of art with gouache watercolours.
What (possible?) protection under copyright law?
In the face of such astonishing advances in artificial intelligence a number of legal issues arise, the first of which is whether the product of artificial intelligence meets the minimum legal requirements of creativity for copyright protection.
As we have seen, since in many cases these are works theoretically comparable to man-made works (which are protected by law on copyright), there would seem to be no impediment to a positive answer to this first question, provided that they are “original” works.
Having said this, however, one must ask oneself in what terms the protectability at the level of copyright of a work deriving from a machine can be conceived, particularly bearing in mind that there is a wide consensus which subordinates the legal protection of a work to the requirement of being the fruit of “human” creative genius.
Other (no less important) issues are therefore who can/should be recognised as the author of such works, and whom does ownership of the property rights to the works created by AI belong to?
An interpretative effort, pending legislative intervention, seems necessary in order to avoid the disruptive effects on the creativity and innovation market that would result from the expectancy that works created by robots could fall into public domain.
Perhaps it might make sense to apply, by revisiting it, the principle whereby the original acquisition of the work realised in the context of a service takes place at the employer’s level and thus to identify the owner of the economic exploitation rights in the programmer or user of the AI system.
Or it can be hypothesised that the ownership of such works be recognised on the basis of an adaptation of the discipline of “collective works” or “composite works”, both characterised by a logic of “splitting” the ownership of the rights, given that, in these hypotheses, the author who holds the moral rights is a different subject from the one who holds the economic exploitation rights.
Otherwise, could a new sui generis “related right” be devised for the producer of works characterised by a minimum level of creativity that are the result of an autonomous AI process?
This is, of course, provided that such provisions manage to fit into the regime provided by our legal system without entailing excessive derogations from the copyright system in force today.
Even more complex is the issue of moral rights on an original work created by an AI: the possibility that this type of rights, by their nature exclusively personal and non-transferable, might be applicable to a machine, appears indeed extremely challenging.
Finally, no less important is the issue of identifying the liable party in the event that the work created by the intelligent machine constitutes plagiarism of a pre-existing work. It would seem reasonable to argue that the person identified as the hypothetical owner of the patrimonial rights arising out the economic exploitation is also financially liable in the event that the intellectual work, created by an AI, is qualified as plagiaristic with respect to a pre-existing artistic invention.
In conclusion, the identification of some of the issues that the emerging digital revolution raises about the nature of the rights at stake and their scope, ownership and possible protection, will undoubtedly require a legislative intervention in the short term, capable of responding to the needs of the subjects involved and able to clarify how and in what terms Artificial Intelligence can in practice constitute a tool to support human creative activity.