Mēs dzīvojam strauju pārmaiņu laikā. Cilvēce ir jau izdomājusi vilcienus bez vadītājiem, virtuālās tastatūras, mākslīgos orgānus, robotus-putekļu sūcējus. Daudzus no šiem jauninājumiem mēs uzskatām par pašsaprotamiem, bet dažus uztveram kā pārsteigumus. Tas, kas mainīja mūsu priekšstatu par mākslu, radošumu, autorību ir mākslīgā intelekta izveidota (uzgleznota?) glezna, kas tika izsolīta Christie's. Šeit ir īss apraksts par to, kā mākslīgais intelekts iesakņojas mākslas tirgū un kādi no tā izriet juridiskie jautājumi*
Mediator, Business Writer, Trademark and Patent attorney
Blurred lines of creation:
Can copyright laws ever catch up with the new world where AI created art gets auctioned at Christie’s?
It is a portrait, after all.
It may not have been painted by a man in a powdered wig,
but it is exactly the kind of artwork we have been selling for 250 years.
(Richard Lloyd, Christie’s)
This October in a way marks the beginning of a new era. For the first time in art history an auction house auctioned out an artwork made by an artificial intelligence (AI). Portrait of Edmond de Belamy created by an algorithm with a lengthy algebraic formula, was put up for 25th October “Prints & Multiples” auction at Christie’s New York (amusingly, in the same auction with works of Duchamps, Miro, Picasso, Twombly, Haring, Lichtenstein, Warhol, Koons and some other well-known names). The description included in the Christie’s Catalogue concluded with a philosophical contemplation: “Time will tell whether the Portrait of Edmond Belamy constitutes a new chapter in this story, or signals the start of very different narrative, where technology is no longer a means to express the creative impulse, but is part of the creative impulse itself.” It might not be the first work ever created by AI, yet this is the first time ever that it was exposed to the market test. The market test was passed successfully with the artwork’s hammer price reaching $350,000 (which with premium resulted in $432,500) and becoming the second most expensive lot in this auction.
Therefore indeed, it might be the start of something completely novel, especially given a myriad of questions this whole situation provokes. However, before I elaborate, let me first sketch out some background for this case.
Thus, a brain-child of Paris-based artist collective Obvious, the generative adversarial network (GAN) as the method is called, created a series of portraits of the fictional Belamy family (owing the name to the AI researcher Ian Goodfellow — in French “bel ami” — who invented the GAN). In its essence the algorithm consisted of two parts: the Generator and the Discriminator. The Generator was creating new images based on 15 thousand portraits fed into it, while the Discriminator was criticising them. This dialogue between the Generator and the Discriminator eventually resulted in an artwork, not resembling anything created before. To note, that this is not anything like a Dinosaur x Flower mash-up by Chris Rodley made with DeepArt.io tool. To stress, Edmond de Belamy is not an AI-assisted artwork, like many others out there; it is an artwork made by AI. In other words, this algebraic formula in the lower-right corner is an author. Or... isn’t it?
In her article “AI Practices and Leadership” Jennifer Sukis subsequently explores the roles of AI as Impersonator, Collaborator and Creator, where AI is assigned a much bigger potential than just being a pimped up tool. It is believed to eventually become a co-Creator and Creator, redefining the whole perception of art and even more — redefining art as such. On the contrary, the artists of Obvious themselves say that: “AI is a new tool, allowing maximization of the creative potential of humans. Nevertheless, for the first time, humans also have the possibility of maximizing the creative potential of their tool.” Therefore, artists behind the algorithm are denying AI the independent creator’s and independent authorship role, pruning it down to a tool with “the maximised creative potential”. Likewise, Obvious in their communication compare the rise of AI art to the advent of photography and video. Back then there was also a lot of criticism and a lot of confusion as to the perception of art. Nonetheless, in a short while the art world accommodated photography and a completely new “breed” of artists emerged — those, who created their art using photo camera.
It could be right: we might regard AI as a tool just like a photo camera; perhaps, more advanced and refined tool, yet — the tool. After all, a human is behind an algorithm. So far. Should AI be the tool, all the questions of authorship and respectively copyright are just to be scratched off as irrelevant. The one who created an algorithm is an artist and is an author in the perception of a copyright law. Full-stop? Not entirely. Even in the current setting, the “tool” created an artwork, not artists of Obvious. Moreover, from the copyright perspective it is not even the artist-craftsman relationship. Artists of Obvious didn’t have a vision on the final product, nor were they involved in the process of creation. Can they be considered as authors? On one hand, sure, who else? On the other...
Let us fantasise a bit, shall we? What if AI is more than a tool? And what if the next more advanced algorithm will not be made by a human? Does it matter, legally? At this moment there is at least an “algorithm argument” to claim authorship. In other words, without an algorithm there would be no Edmond de Belamy, so the one who made an algorithm and set it in motion with a vision to create (even without a vision as to the outcome of creation) — is an author and should have a copyright over the final work. But what if AI is ever behind both its algorithm and the result of the algorithm put to work? Should it then be granted copyright? Here we come into the clash with the very essence of our current intellectual property (IP) laws. The rationale behind current IP laws is preventing others from using IP objects and enabling the IP owner to get a benefit. Out of this logic, can AI be interested in benefit? No, at least not in our, human, perception of it. Another classical legal consideration is: whom then to sue for violation? Obviously you cannot sue AI. In other words, legally speaking, we are completely in the dark here and the whole notion of IP and its applicability (if any, if ever) to AI has to be conceptualised, before we can dream of ever applying it.
To add an anecdote: recently there was already a copyright case, where non-human subject was claimed to be an author. A Naruto case in San Francisco — a monkey who took a selfie with a British photographer’s camera. The US Copyright Office in this case concluded that human authorship is a requirement for copyright protection, and therefore a monkey cannot own photographs. This conclusion was also confirmed by a court. Therefore in the current legal framework, if a monkey cannot be recognised as an author, that doesn’t give much hope to AI either. The question, however, remains: with such a speed of developments in the field of AI and with its confidently entering into the art world, will our copyright laws ever catch up? Will we?..