Redefining Creativity: From an Idea to Representation and Back
Updated: 6 days ago
Last couple of weeks have been filled with variety of information, news and anecdotes related to artificial intelligence and in particular, to its last incarnation – chatGPT. The commentators, bloggers, Youtubers and other content creators have written or chatted about this and that chatGPT can do on numerous accounts. Myself included. I have also written series of articles dealing generally with chatGPT , its impact on legal profession  and about DALLE-2 and generative art .
Meditations on creativity
In this piece I want to delve into the closely related topic, that of a creativity. It has been quite a surprise for me to see that the activities which were generally considered creative have actually been first to succumb to the potential of the AI. DALLE-2 and other tools can now generate art on demand. The results are second to none.
When thinking about it on a deeper level, have the machines become creative? Or, since a prompt from a human operator is still required in order to instruct the machine to generate the outcome, is it still the human on whose creativity the result is very much dependent?
A copyright perspective
The current Czech law provides that the copyright-protected work is a „unique outcome of the creative activity of the author […] expressed in any objectively perceivable manner“  while at the same time, an author is a natural person who has created the work . Thus when it comes to copyright, first, a result expressed in any objectively perceivable form is protected rather than the idea and, second, only humans can be authors.
That being said, it follows that if I were to request DALLE-2 to give me a „an impressionist painting of the juxtaposition of the uptight lawyer surfing the big waves in Hawaii inspired by works of Rene Magritte”, the result is what would be protected under the copyright, not the idea.
But can this picture, in this case, be truly considered a unique outcome of my creative activity? Sure, I made up the initial prompt, which could be considered a creative activity, but would it not also be just an idea? My point being, to what extent am I the one who not only conceived the idea of the picture, but also made the work that resulted in the outcome eligible for copyright protection? The AI is the tool here but still, the degree of its independence is much greater than in case of some other tools (both digital and analogue).
It follows that if I were to be considered author of the above impressionist surfing lawyer Magritte-like painting, would not the copyright protection in that instance somewhat shift towards protecting the idea rather than its expression, since the degree to which the painting is really an outcome of my creative activity is somewhat debatable?
However, apart from the conclusion (1) in which only I am the author, and thus at the same time conceding that copyright is perhaps protecting more of an idea rather than the expression, there are basically two other options to consider. Either (2) both me and DALLE-2 are authors, or (3) the above picture is not eligible for copyright protection.
Is creativity an ability bound to be reserved for humans?
Where does that leaves us? Could machines ever become authors? Well, certainly not under the current legislation, since as seen above, only humans can be authors.
However, setting this aside, as laws can be changed after all, is creativity truly something only humans can perform? In case that we strive to create a true artificial intelligence, it should very much possibly be able to perform also intellectual activity on its own.
It follows that if the creativity is a function of intelligence, the AI should then be perfectly capable of being creative as well. Hence, it should be possible for it to become author since the condition that the author must be human in all circumstances would no longer make sense, or at least would not have any logical justification.
Perhaps the only way of how to be able to maintain that creativity is truly something only humans are capable of, or that it at least should be something reserved for them, would be to somehow link the concept to the idea of consciousness. For example, by positing that only a conscious beings can truly be creative and thus authors of works capable of copyright protection.
This, of course, opens an entirely different but no less fascinating topic – that of the nature of consciousness. Even if we were to accept the above – that only conscious beings, such as humans, have what it takes to become authors – what if the AI would claim that it has become conscious as well?
Circling back to the three options of how to approach the question of the copyright regime of the AI generated art, as described in paragraphs above, I would favour the second option. That is, both me and the AI could be the authors.
My subtle proposal is that the copyright, in such a case, be viewed more as an industrial right similar to patents, trademarks and designs, rather than as it is viewed now – concept linked directly to the artsy side of the creation, closely intertwined with the personality of the author.
Much like the patents have their inventors, who are most of the times a natural person other than the entity exercising the economic rights resulting from the patent, also copyright-protected works could have their authors as well as proprietors holding the economic rights. It is not exactly a revolutionary idea in any sense as this is more or less already the case in Anglo-American concept of copyright, and even in the continental Europe it is very often the case with respect to software and databases. But still, I think that we could benefit from a clearer and more distinct shift towards copyright as an industrial right.
That way it would be possible to differentiate on one side between author, in the sense of who got the idea and potentially who also contributed to the creative activity, while allowing for the machines to become authors in this sense as well. And on the other side between the one who actually holds and exercises the corresponding rights. As already pointed out, this is often the case with software and databases, so what harm would be in further formalisation of such regime also for the rest of copyright-protected works?
The generative art created as a special blend of human ideas, which would not themselves be eligible for copyright protection, and machines’ creative activity, could become a legal category in its own right.