Numerous articles discussing the advancements of Artificial Intelligence often criticize it for “stealing” human ideas and creations. In this article, Elvin Blakaj challenges this viewpoint.
The proliferation of articles on AI’s progress in every aspect of human life is becoming increasingly prevalent, sparking global discussions that are now reaching Kosovo. Particularly, the impact of AI on visual arts has become a topic of intense debate, as the technology rapidly infiltrates a domain traditionally associated with human creativity.
how AI-generated art should be treated compared to art created by a human being?
As a society we see AI as superior in specific fields — generally technical ones — but the fact that such technology is penetrating and rapidly advancing within a field we believed to be exclusively human — art — creates a new need to redefine AI. Not just as to how and why AI can create art, but even further, to whom exactly then the notions of imagination, style and soul should be attributed to now?
One recurring question stands out in the debate about AI and art: how AI-generated art should be treated compared to art created by a human being?
In the plethora of contentions with which I agree with these articles argue that what makes art, art is the presence of a distinctive intention that flows from within the person who creates it. In other words, art is art only when there is a mind behind it that, through the creation of the artwork, expresses something about the world around them – internal or external.
But there are other claims within this debate I think have less merit! An article published in Kosovo 2.0 by Donart Zymberi illustrates one such argument. He argues that simply prompting the AI to create art in the form of Van Gogh would be stealing from him. I don’t believe this to be a good assumption and I’ll illustrate why.
One point where I disagreed with this argument is the evaluation that generative AI, through the creation of art, is not merely generating something but is stealing from artists, whose art it uses to learn from. In other words, the claim used here is that AI steals from the artist to create its own art – or to generate it if we use the correct technical concept.
To clearly understand why I think this argument is wrong in terms of its conceptual basis, we need to clarify two notions: what we mean by “learning” within this debate, and how we see AI within all of this.
There are hundreds of definitions of what learning actually is, but to summarize it in one sentence – not absolute but general – learning can be conceptualized as the process of absorbing past information for some effect in the future.
It may seem very simple, but practically every technical definition has this essence in its own practicality, as mentioned earlier. And while we could delve infinitely into this, for the purpose of this opinion, the definition in question is more than sufficient, as we clearly see that it lacks something – the medium!
Humans are not the only entities capable of learning; animals and even machines, have the capacity to learn in their own way, which puts us on a crossroad regarding the meaning we want to give to ‘the process of learning’. Only one of these two sentences can be true: every learning is theft, or none is.
The artist learns form by copying the art of the greatest, would say Walter Benjamin, a cultural essayist of the 20th century. Anyone who creates any form of visual art can confirm that fact because to make art, you must understand its essence, and that is achieved through practical immersion in the artists of the form you wish to understand.
The problem is that deep learning AI does literally the same thing; it takes a tabula rasa system [clean of knowledge] and through immersion in thousands and thousands of art points, it learns what is, for example, a painting of Picasso as opposed to one of Dali.
It learns how M.C. Escher creates woodcutting in this particular style, while Baumann does not? The learning process of AI is fundamentally based on learning from mistakes and understanding from the past, which is entirely identical to how a person learns art [or anything else, really].
As much as I understand the anguish that artists feel when they see their works, which took decades to be learned and improved, being replicated within a minute, the fact remains that the argument presented for learning and theft is counterproductive because it can also be used against the artists themselves.
Every artist, in one way or another, has learned and continues to learn from the past forms of art that have been created. Therefore, in the words of the legendary artist Picasso that art begins with copying, some only evolve in more significant directions than others.
These types of arguments misconstrue the reality of learning through the process and attempt to separate the form of learning from the medium that shapes it.
Such arguments undermine the entirety of learning as a basic concept of human existence. And why is that? Because we all are fearing something that seems to possess the spark of imagination we thought was exclusively ours, but we need to learn to share it because it was never solely ours, to begin with.
Machines of skin and flesh are ultimately just one kind of the many learning-machines, a fact that we must slowly come to understand.
This generation of AI is the weakest version we will ever see, and if it frightens us so much today, then the next ones will terrify us tomorrow as the evolution of these machines is marching forward every single day and we need to learn how to understand – and deal – with what is to come appropriately.
Elvin Blakaj is a research scholar with a primary focus on international relations and international law. He is currently working as a peer reviewer and research assistant, and his PhD in International Law from the University of Sydney. His interests lay in philosophy, history and anthropology.