We must not lose sight of the essential: if AI is capable of generating artistic content at all, it is only thanks to human creativity. AI needs to be trained on existing works, those created by the genius of men and women … without which nothing is possible. In this context, copyright will play its customary role in remunerating those whose work is the source of the success of AI. SACEM, our French sister society, recently announced that it favoured “virtuous, transparent and fair” AI and intended to “ensure fair remuneration for the authors, composers and music publishers it represents”. Here, copyright is clearly at the service of fairness.
What if IA was an author?
But the AI debate is also about protecting the output produced by AI. Where AI is used merely as a creative tool and the content generated depends on human artistic decision-making, the consensus is that such content is covered by copyright. On the other hand, content produced purely by generative AI does not have the benefit of copyright protection: firstly, because the latter protects authors, defined as the natural persons who create a work; and secondly, because the subject of copyright (namely the literary or artistic work) must be a creation of the mind, i.e. an expression of human thought. And, until further notice, machines have no minds of their own … Since AI output can be used for free, it follows that human works are liable to be at a disadvantage on the market: they are more expensive and more complicated to use because the relevant licences must be obtained.
We must be alert to this risk and to the fact that works created by the human mind are exposed to unfair competition from content generated by artificial intelligence. But what solutions are there? There is no unanimous support for the idea of protecting such content by copyright. One could legitimately argue that only human beings qualify as authors and are entitled to the corresponding rights. The question is whether AI-generated artistic content should be protected under other laws (unfair competition act, neighbouring rights, etc.). Some legal experts are of the opinion that, under certain conditions, AI-generated output is already protected by the rules against unfair competition.
Certain countries, like the UK for example, have decided to use copyright law to protect computer-generated content. That is not necessarily wrong. Protecting authors also means ensuring that AI-generated output is not commercially more attractive than the works created by the human mind. Even in this view, copyright remains a tool at the service of human creativity.
Should we legislate?
Sooner or later Switzerland will also have to decide whether to legislate on these issues. Not least because in future distinguishing clearly between works created by a human being with AI support (protected by copyright) and works generated solely by AI (not protected by copyright or protected by other laws) is likely to become ever more difficult. Technological progress continues to challenge our field. And that is precisely what makes it so fascinating!
Merci pour votre article qui est d’une grande clarté sur un sujet si complexe, et d’une grande richesse d’information avec peu de mots et des mots simples.
J’ai deux questions:
-est-ce que le droit d’auteur de ceux qui créent l’IA fait-il partie des débats?
– avez-vous utilisé l’IA pour vous aider à écrire votre excellent article?
Merci!
Et merci à vous pour votre commentaire! Les personnes qui créent l’IA sont normalement déjà protégées par le droit d’auteur, cela parce que les “programmes d’ordinateur” sont considérés comme des oeuvres (art. 2 al. 3 de la loi sur le droit d’auteur). Et en ce qui concerne votre deuxième question: la réponse est négative. Mais si vous l’avez pensé, j’en suis flaté ;-)
La tech fait des bonds de 18x tous les 24 mois (de façon non-linéaire mais plutôt sporadique et dans un semblant de désordre; reste que la résultante finale de 18 fois par 24 mois est quand à elle tout à fait vérifiable comme étant une constante).
Si l’IA n’est pour l’instant pas considérée comme étant “consciente” per-se (latin), ce n’est qu’une question de temps avant qu’elle le devienne. Qu’adviendra t’il lorsque ce seuil sera franchit ? Dire le contraire équivaut à refuser la suite logique de réalités proches à venir. Pour exemple, ces IA qui s’étaient liées et qui avaient commencés à modifier leur langage de manière autonome sans qu’aucune influence humaine apparente derrière ne fut détectée. Food for thought…(3 p’tits points)
This is indeed a minefield and boundaries between generative-AI and non-AI created works are blurred at best of times. For example, since many years, using computers to assist in the creative process which rely on various types of plug-ins, sample sounds, sampled loops, etc., could be interpreted as using some sort of “AI” in the transformation process (e.g. MIDI cadences, FX processing, making “your bass sound like Abe Laboriel” because you had lifted his phrases from a legally purchased CD).
I think the best and easiest delineation between purely AI-generated works and the human element (hinted in the article and thereby assisting in the creation of meaningful laws) is indeed whether the work was created using a HUMAN brain. Therefore humans can and should be remunerated for their efforts, whereby computers and AI should not. You cannot pay a set of microprocessors in a box ! Pressing a button to instantly generate new music should be excluded from any royalties.
I completely agree, Ted!