Forum

Artistic singularity and artificial intelligence

Edorta Ruiz

4 July 2018
13 comments
Total votes: 3

Hello everyone. My name is Edorta and I am writing from Spain to explain my doubts about artificial intelligence. I dedicate myself to painting, and luckily it is my source of income, besides being my great passion in this life, I do not do anything else but paint. I have been reading for several months about the impact of artificial intelligence in the art world and I have several doubts about it; Is it necessary to use machine learning to carry out the work of a painter? Painting is something that comes from the person and I do not see any use for this purpose. It could be used to streamline processes, but in my work, for example, the process is very important. I have read that the software that many companies are working on is capable of generating a large number of images with just one click. My question is this: Are not you risking devaluing classical art? Can not the companies behind these creative algorithms make a huge profit? Is not the creative uniqueness of each artist put at risk? I think this is something to keep in mind when developing such algorithms. Imagine that one of my works coincidentally looks a lot like one, or is exactly the same as that generated by an artificial intelligence. What would happen here? Would the person's work be discarded even knowing that it was something totally improvised? Please provide your comments as it is something to take into account. Thank you.

Tags:

art property law singularity

Comments

Felix Viberg,

10/07/2018 17:27

Excellent question Edorta.

I, too, find the question of artistic expressions very interesting and wonder what it will mean if art (of different forms) can be mass-produced without much labor, and I tend to end up at similar, not firm, conclusions each time i reflect around the meaning of the question.

For me art of all forms must always convey some sort of message whichever domain it occupies, may it be dance, music, painted art or how to present a dish at a restaurant. I believe that all art forms is merely a medium of expression and thus, somewhat protected from artificial entry.

Take painted realism (naturalism), I find it likely there existed concerns that this form of art would become outrivaled after mankind invented ways to project imagery from the 3D world into a 2D surface.. But whenever I see realism in arts today my mind directly reflects upon the achievement that the painting is - "how is this level of realism possible?", "why did the artist choose this picture?", "is it possible for me to do something similar?". Of course, the invention of cameras does not answer those questions, instead cameras created a bunch of new art forms.

I have a hypothesis that I would like you all to try to dissect.

A near-fundamental question that arises whenever someone is a receiver of most art is - "is it possible for me to understand this?". And in the case of AI-produced (replica-)arts the answer is "no" as long as we do not believe the AI is self-conscious. Thus the question is completely protected - as long as you know the art is man made.

If there was a way to deduce that the hypothesis would hold, what would that mean?

Richard Krajčoviech,

10/07/2018 16:48

Hi, Edorta,

my view is that Artificial intelligence is just another tool available to artists. I think that painters went through same dilemma when photography has been invented. There were fears that photography will replace painters. Did it? To some extent yes - those who specialised in painting portraits, realistic land views, etc. But photography is limited to capturing reality (with some artistic filters). Finally painters realized that realitic puictures are just tiny portion of what can be painted and that there are plenty of themes and styles that cannot be done by photography. And even classical art still keeps its value.

Artificial Intelligence might affect some painters, but I am sure that the creative ones will either find area that cannot be done by Artificial Intelligence or utilize Artificial Intelligence for creating their specific style, as likely as phothographers express through photography.

Franc Paul,

10/07/2018 15:50

My wife is also an artist, a painter. I appreciate art for many reasons. Amongst them are that I value beauty, skill, use of analogy and methaphor, and the communication of something that it is like to be a person.

At some point non-human intelligences will be able to do some or all of that; and I will appreciate it.

There will be more and more humans creating works of art to convey what it is like to be a human. Other humans, and intelligences, will value that.

We are on the cusp of a great creative era where more and more people will have the opportunity to express themselves creatively. That is if the promise of AI's abundance is managed for the benefit of all.

Right now someone can take a good quality photo of your work, and have a good quality print made of your work. Something is lost in this translation, and the print is obviously not worth the same as the original. The provenance of a piece will always play a large part of the value of it. If your story is compelling you shouldn't worry about an AI happening on the same style, if anything it will make your story more interesting.

In the near term, there are many more profitable uses of time spent developing an AI that copies the style of a living artist and pretending the works to be an original by the artist. The provenance of the first artwork created by an algorithm that is judged to be creative will ensure that that piece is valuable. The value curve for subsequent pieces by the algorithm might decline sharply. or not.

Jean-Marc Deltorn,

07/07/2018 10:35

Dear Edorta,

Thanks for this post: it raises important questions related to copyrights and automated productions (e.g. using AI tools).

As a researcher working in the field of AI and intellectual property rights (including copyrights), I could help clarify some of your doubts.

First, note that the the interplay between computer-based creations and IP rights is far from being novel. Even though the recent advent of a new branch of generative tools (e.g. based on deep learning processes) gave this rather old debate a new impulse, many of the issues have already been discussed either at a doctrinal level or in court (and we should certainly draw from this experience).

I have addressed these issues in a paper:

J.M. Deltorn “Deep creations : Intellectual Property and the Automata”, Frontiers in Digital Humanities, vol. 4, n° 3, p. 1-13, fév. 2017.
accessible here: https://www.frontiersin.org/articles/10.3389/fdigh.2017.00003/full#

I am quoting/paraphrasing below some passages of this text relevant to the present discussion:

First, to be copyrightable a work must be "original": in the vast majority of jurisdictions, this notion establishes a particular link between a human creator and a materialized object. Indeed, the presence of the author’s “personality” in the created form is a cornerstone of the modern principle of artwork copyrightability and is, symmetrically, an essential component that defines the author.

That a creator is required for an artwork to be copyrighted appears, therefore, as a legal imperative. In most Continental European copyright systems, the creator is considered the de facto author of the work. Furthermore, only a natural person may be considered an author: the French “droit d’auteur” refers to the copyrighted work as a “creation of the mind” (“une oeuvre de l’esprit,” as defined in art. L.111-1 of the French Intellectual Property Code). Similarly, section 7 of the German Copyright act states that “the rights holder is the creator of the work” (“Urheber ist der Schöpfer des Werkes”).

The requirement of a human author has also been made explicit in U.S. copyright law in the context of artwork generated through automated processes: the U.S. Copyright Office has indeed now taken the position that “in order to be entitled to copyright registration, a work must be the product of human authorship. Works produced by mechanical processes or random selection without any contribution by a human author are not registrable.”

So, in a nutshell, in these frameworks: no human author -> no copyright.

Works created by a machine, without human (original) intervention should be considered, for all practical purposes, in the public domain.

Still, in a few jurisdictions (UK, notably), consider that, when the creation is automatically generated “the author shall be taken to be the person by whom the arrangements necessary for the creation of the work are undertaken.” (section 9(3) of the UK copyright code). While this undoubtedly prevents a computer-generated creation to fall in the public domain, the law leaves nonetheless undefined, and therefore open to interpretation, the exact role of the person by whom the arrangements are made...

This highlights - in passing - the importance of proper definitions. The idea of the "machine-as-an-artist", is a rather elusive concept, largely ill-defined (even in the legal literature) and often extrapolated upon (as other concepts related to "singularities" or "general AI", that do not correspond to any technical reality, but that obviously serve to raise the hype...). As a scientist I could easily argue that there is always a human in the production pipeline (the user of the deep neural network, the programmer who implemented it, the individual who selected the training set on which the internal weights of the DNN) and that, for all practical matters, the machine-creation do not happen ex nihilo! However this human presence is neuther systematically nor necessarily "creative/original" and in this sense should not justify the attribution of a copyright without a detailed analysis.

Finding the author, however, may not be so trivial when the intervention of an automaton in the creative pipeline, in the form of a non-linear algorithmic process, blurs the human contribution to the point where it is hardly discernable anymore. But, again, this is a matter that should be judged on the merit, i.e. on a case by case basis. Generic answers (including those vague statements about "machine creativity", i.e. considering the "machine as being creative") mostly fail to address the practical points and leave the community in doubt as to the correct interpretation of the legal norms.

If copyright is to remain an incentive to creativity, an adaptation of some of its underlying principles may be required. Allowing for a fair use of existing artistic corpora while limiting the attribution of rights to those artworks that reflect an original contribution of human creators might help to maintain the function of copyright as a catalyst to creativity.

Edorta Ruiz,

09/07/2018 09:53

Thanks for your answer. It has helped me. In the case of painting, I think that personality is something that is reflected in each work of the author / painter, so it will always be something unique and unrepeatable. I think that artificial intelligence can be used as a tool, but never as a substitute for our way of expressing ourselves. Thank you very much and greetings.

Jarosław Dukat,

09/07/2018 19:10

Edorta, however sad and disturbing it must sound to you, the style is repeatable. Check this:
https://www.google.com/search?q=neural+style+transfer+examples&safe=off&client=firefox-b&source=lnms&tbm=isch&sa=X&ved=0ahUKEwjY__-MwZLcAhXN6qQKHdq3CAUQ_AUICigB&biw=1920&bih=953

It is called neural style transfer. Very interesting deep learning based technique.

Cheers,

Jarek

10/07/2018 13:32

It will be inevitable that many artists refuse to use artificial intelligence for their works, since we can not force anyone to use it. As for the transfer of style is used to replicate great artists already dead, not to simulate living artists, because that would be intolerable, I would at least claim it judicially. Each painter has the right to be unique in his painting, besides that it would be intolerable since the artistic singularity would be lost. That would be the death of humanity.

13/07/2018 10:24

I fully agree: "AI" is a nothing but a tool for the artist and should not be assimilated (as is often the case in the media) as an independent entity capable of intentionality. There is no such thing as and "AI personality" in the sense we define it in copyright law (what has been debated, e.g. at the EU level in the "recommendations on robotics" regarding the possibility of an AI personality is based on very different premises, mostly intended to address issues related to responsability/liability, and could only (at best) be seen as a limited/reduced personality related to traceability).

As such, we may witness a cornucopia of novel computer-based intermediaries that will provide new forms of creative opportunities for human artists, new ways to express, as you rightly point out, their personalities. This is the framework in which copyright law should be delineated (which will certainly require serious and challenging work, as is often the case with the introduction of new technical tools or new artistic practices).

As a side note: I believe that, as legal practitionner it is our responsibility to remain as precise as possible and avoid jumping to conclusions over ill-defined notions ("AI" is in this sense a rather good example of such a notion in dire need for precision...) and avoid falling in the trap of fads or hype, how tempting and exciting these may be. To do so it is essential to draw a proper technical understanding of the object of interest (namely, the rather elusive concept of "AI"). This mandates, I believe, a multidisciplinary approach: where technical experts (e.g. computyer scientists), legal practitioners, artists, and others, exchange honestly and openly.

I hope this forum will provide such an opportunity and I thank you again for lauching the discussion!

09/07/2018 19:07

Jean-Marc, you provide very valid arguments and it all makes sense in general.

Yet it is still difficult for me to agree with you fully. A painter uses tools: brushes, canvas, paint and using these tools and expressing himself, creates the artwork in a complex, time-consuming process of creation. It is copyrightable.

A photographer uses a digital camera, and with one click expresses himself, taking a photo. Very easy process. Of course, taking a masterpiece photo is difficult. Yet the process itself is quite trivial except it is not possible without the camera. It is copyrightable, though.

A software engineer creates a programme. It is just a sentence of letters. Not possible without a computer, though, etc, etc, and copyrightable.

So, why wouldn't a new generation of artist train neural network and then click it, like the photographer clicks his camera, and create artwork?

Similarly, with the neural style transfer, the network is trained to mimic the style of some picture and apply it to another one, but it is the human that picks and combines both pictures. Why wouldn't that be copyrightable? There was certainly a human factor there.

Of course, picking up pictures can be automated and applied randomly to generate a lot of pictures.

I'm just trying to say that the mere fact of using a neural network to generate a picture doesn't automatically imply it is not art.

13/07/2018 10:35

Dear Jarek,

I actually fully agree with you (I might not have been clear in my previous message, if so, I apologize for the confusion).

There is indeed absolutely no reason to deny copyright to an individual on a artefact produced through the intermediation of a computer tool, provided the product is original (in the sense of copyright/author's rights) and entails specific choices by the artist.

That these choices emanate from (e.g.) picking parameters in a deep neural net, from selecting a training corpus, or from interacting with an algorithmic engine does not really matter. What matters is that the "personality" of the artist is expressed in the tangible ("fixed") expression that is eventually produced (as you rightly point out, the issue of the effort involved in the creation is of no consequence).

However (just a caveat) this is not systematic and should be (as always) examined on the merit.

I develop these issues in the paper I referred to in my response to Edorta's post (in case you feel like digging further).

Kind regards,

Jean-Marc

Bjoern Juretzki,

19/07/2018 15:23

Dear Jean-Marc,

I fully agree with this excellent overview of the problem. You rightly point out that it is difficult or even outright impossible to copyright works that are produced by computers. But are companies that employ algorithms for journalism, music production, etc. fully aware of these limitations? Will they not push at one point for a change in rules to make also AI-produced works copyrightable?

Especially if societies decide at one point to clearly identify/mark whether something is AI-powered or not, people may quickly exploit this copyright loophole.

Elio PENNISI,

05/07/2018 09:37

Edorta, to answer your dilemma, I would be tempted to apply the same principle governing creative work within research departments of companies. The employee (researcher) as he/she is recruited is asked to sign a document stating that all outcome (findings, products, processes, etc.) derived by work (and instruments) effected inside the company, is the copyright of the company and not of the individual researcher.

As AI systems are certainly owned by a company or an independent individual, mutatis mutandi, the/any artwork or outcome produced/derived by an AI system becomes the property (copyright) of the company or the independent AI-system-owner.

Note: reactions to what expressed above will be much welcome..

Juergen HEILES,

04/07/2018 15:34

Hello Edorta,

you raise interesting aspects. For me creativity is a major aspect of art. It is your creativity that leads to your paintings. Just painting a one to one copy of a photo is not creative for me and therefore not art. The photo itself reflects the creativity of the photographer, his selection of the scene, the exposure and so on. However creativity is not limited to art, an engineer that designs a new machine and an architect that constructs a new house use their creativity in the design process. Carpenters, plumbers, gardeners they all use their creativity to handle a specific case and come up may be with a unique solution.

So is AI creative? I don't have a final answer. If you train an AI with paintings from Rubens it will paint new pictures in the style of Rubens. Is that creative? But you may have an AI that is not trained by using existing pictures but learns by its own (reinforcement learning) like AlphaGoZero for playing Go. Are the moves creative? A game usually has a defined outcome (win or loss), what would be the desired outcome a reinforcement learning painting AI would be trained for? How much unique would that be? Isn't the definition of the training goal in that case the creativity behind the AI? And that would still be a human.

But coming back to your questions on devaluing art, profits and uniqueness. They apply to any (creative) work in my view. Is the engineering and architecting work devalued in case AIs would perform the task? Can solutions that are generated by an AI be patented? What would happen if a human or another AI comes up with the same solution?

Especially for art the questions is what defines its value? Would somebody pay as much for an AI painting as for a human painting? I don't have answers for that. Can we avoid that? No, if it can be done it will be and if AI paintings are valued and bought by people it will be a business. We have to see how much uniqueness humans can still have compared to AIs.

So no real answers for you and more questions. We are just at the starting point of AI and I assume that we will have even more questions and only limited answers.

Juergen

European AI Alliance

ARCHIVED PLATFORM

This website is archived since mid-May 2021 and it is not possible to login anymore.

But don't worry there is a new Futurium platform and all active communities have been migrated.

Forum

Artistic singularity and artificial intelligence

Comments

ARCHIVED PLATFORM

This website is archived since mid-May 2021 and it is not possible to login anymore.

But don't worry there is a new Futurium platform and all active communities have been migrated.

Share this:

Comments