In recent years we are witnessing the confluence of great technological advances that due to the speed with which they are taking place, the breadth of sectors of the economy and facets of our lives that they are affecting and the profound disruptive impact they are having, It has led experts to affirm that we are facing a new Industrial Revolution, the fourth, which will lead to a transformation of humanity because it will fundamentally affect the way we live, work, and interact with each other.
Many of the changes associated with this revolution have to do with advances in the field of artificial intelligence (AI), that is, with the emergence of machines and computer systems capable of emulating the human mind. As the press is responsible for reminding us almost daily, for certain tasks, these AI systems are even more efficient than the human mind: diagnose skin diseases; play chess, Go, poker or video games; analyze legal documents, or translate texts into a plurality of foreign languages.
Table of Contents
Machines that create art
These machines are not only very useful for a man when it comes to producing artistic creations, but many of them are capable, on their own, of generating those creations. Thus, the members of The Next Rembrandt project designed an algorithm that, from the analysis of all the works of the Dutch painter, generated a new painting by himself imitating his style.
Another example is found in the picture by Edmond de Belamy, attributed to the Obvious collective – three French students who, by the way, obtained $ 432,500, about 395,000 euros, for the auction of the work -, and generated by the machine learning algorithm GAN, whose ownership belonged to a researcher named Ian Goodfellow.
To do this, the three students trained the algorithm with more than 15,000 pictorial works in the public domain. The press also tells us about algorithms capable of writing news based on brief facts, poems or even scripts of television series.
The emergence of these creations generated autonomously by machines is leading experts to wonder if they should receive the same treatment as works generated by humans. That is if these creations are subject to copyright protection and, where appropriate, to whom the ownership of those rights should be attributed.
The simplest answer is to affirm that these creations are susceptible to the protection and that the ownership of the rights belongs to the person – physical or legal – or group of people behind the machine. It is the solution offered, for example, in UK law, one of the few that offers explicit regulation of the issue.
The answer, however, presents problems that highlight the ethical and philosophical connotations that the regulation of any aspect related to artificial intelligence has.
In the first place, in many cases, the identification of the person behind the machine, that is, the one who carries out the necessary arrangements for the machine to generate the new work can be problematic. Thus, for example, in the case of Edmond de Belamy’s picture, this person could be both the creator of the algorithm and the three students who compiled and digitized the charts with which the algorithm was trained.
The question is of great importance because, if the developer could be considered an author, he would be entitled to a part of the price obtained by the auction of the work. Unfortunately for Ian Goodfellow, in this case, it is easy to conclude that the authorship corresponds to the three French students because it was these who really contributed to the creation of the work.
Complex task due to excess and default
On other occasions, this task can be more complicated by excess – there are many people who have contributed to the creation of the work – or by default – none of the people involved has made a contribution that can be considered essential for the generation of the work by the machine.
An example of the second assumption would be that of a journalist who merely introduces a series of facts from which the machine conducts a search for information in its own database and on the Internet and generates a newspaper article.
Is it fair to attribute the authorship on this article to the journalist? Doing so would distort the copyright system, based on attributing to the author a right of exclusivity on his work as a reward for the intellectual effort made for its creation, from which the whole society benefits.
In those situations in which the person behind the machine does not make an essential contribution to the generation of the work, it should not be considered an author. To achieve this result one could speak of the existence of a general rule, the attribution of the ownership of the rights to the person behind the machine; accompanied by an exception applies in those cases in which said person does not make an essential contribution to its creation.
The problem is that this solution can only be provisional because, as advances in artificial intelligence favor the automation of machines, the exception will gradually tend to become the general rule. At that point, can we continue to state that the solution is consistent with the basis of the copyright system?
A second possibility, rejected by practically all specialized doctrine, is to attribute the ownership of the work generated by a machine to the device itself. Currently, with the law in hand, this solution is unfeasible. Although international texts do not explicitly say so, many of its provisions – for example, the one that determines protection in response to the author’s nationality – implicitly require that the author be a human being.
For the current legal system, intellectual creations susceptible to copyright protection can only be produced by the human mind. This, however, implies assuming that artificial intelligence cannot be assimilated to human intelligence – since only the creations of the second are susceptible to protection – something that can be shocking to those who are sure that technological advances will allow developing systems. of general AI — capable of performing any intellectual work that a person would carry out — and that, as Ray Kurzweil said: The Singularity is near.
In view of these more progressive approaches, a modification of the law could be postulated to attribute ownership of the rights to the machine. This solution, however, would pose serious problems.
In the first place, because that would imply providing legal personality machines, a decision that would have implications beyond copyright, since it would require changes, among others, in the field of civil, criminal and financial law.
Secondly, because attributing the ownership of the rights to the machine does not respond to the purpose for which the protection system was created: a machine does not need to be rewarded with an exclusive right over its creation to incentivize it to produce it.
And with this, we arrive at the last alternative: as the works generated autonomously by the machines cannot be original, they are not susceptible to protection, so they are in the public domain and can be exploited by anyone with access to the work. This solution has the advantage of favoring the general interest because such actions can be used in research projects, or as educational materials, or for the simple enjoyment of the average citizen.
In any case, it is a problem-free solution. To begin with, it is complicated to identify which works have been generated exclusively by machines. The user of the machine can be attributed to the authorship of the work and, contrary to what happens to the cheats who resort to a literary black, there is no danger of being denounced for plagiarism – the machine is not going to goat. In the matter of patents, it has been denounced that countless patents requested and granted by the US Office to natural persons were conceived and developed by machines.
Another problem presented by this solution lies in the indirect damage that natural persons may cause to authors. That there is no copyright protection for works generated by machines means that the costs for access to these productions are lower than those for works produced by humans, which, being protected, making it necessary to negotiate a price for authorization to its exploitation
For some experts, in the long run, this could lead the public to acquire works generated by machines to the detriment of human-generated works. This harms the market position of the works of authors natural persons, whose prices should be reduced; and would discourage human intellectual creation.
Some experts minimize these risks because, in their opinion, the creativity of the machines is limited, and at no time will they be able to match human creativity. In his opinion, the public will always be willing to pay a higher price to access human creations. Again, these arguments involve assuming a very conservative stance on the advances that can be reached in artificial intelligence.
Today it is possible to affirm that a newspaper article about a specific event written by a machine lacks the depth of reflection of an opinion column about that event. However, is it possible to rule out in the future a scenario like the one presented by Max Tegmark in Vida 3.0 in which, Prometheus, the AI system designed by the omega, designs the script of a television series after analyzing the information available on the Internet about the tastes and hobbies of the world population? Is it possible to affirm that this script cannot have the same creative height as one written by a human being?
As can be seen, none of the three answers proposed on the attribution of copyright to creations generated by machines provide a definitive and problem-free solution. Perhaps the correct answer is that it cannot be, or rather, definitive solutions should not be adopted.
The IV Industrial Revolution also affects the legal system. At present, flexible legal norms are needed that can be adapted to the vertiginous changes that are derived from the exponential technological growth that we are attending. Thus, in principle, the ownership of the copyright on the works generated by machines could be attributed to the person behind the machine, with the exception of those cases in which there is no person who has contributed substantially to the production of the work.
In such cases, it should be understood that the work will not be protected and will be part of the public domain. Advances in artificial intelligence can cause the application of this exception to be increased and, even if these advances lead us to admit that intelligence is not exclusive to humans, the machines could become authors and even subject to rights.
The future offers very interesting challenges to jurists for whom we must be properly prepared if we want to provide appropriate solutions to the demands of society.