Francesco d’Isa on AI art: data, errors and kitsch

Acronyms:

FI - Francesco D’Isa

AS - Anastasiia Spirenkova

AS: In your work for the TAEX exhibition Scoletta dell’Arte: Digital Reform, curated by Antonio Geusa, you raise the topic of the aesthetic beauty of errors. In the course of creating this artwork you even taught artificial intelligence how to make mistakes. How did you come up with this idea? Is the pursuit of perfection a bad thing?

FI: Let's say that AI in its production of images often repeats biases and prejudices and therefore provides very stereotyped images. When I work on it, I find it more interesting when the machine is ‘taken away’ from this. Normally, AI has this tendency to create average, mediocre images that are more stereotyped, even kitsch. By forcing a machine to make mistakes I prefer to detect its potential creativity, novelties and elements of surprise or interest.

So let's actually say that it's a ‘malfunction’ of the machine that is able to return images that are not exactly the most predetermined, but more unusual and where, from an artistic point of view, much more material of interest can be discovered. It also happens in art in general, when through a mistake you can distinguish the new potential ideas.

I find it to be a part of creative thinking. Errors are invaluable for an artist as they lead to new perspectives and unveil the freedom of thought. If you think about it, all new visions were mistakes at the beginning.

Francesco D'Isa, Errors, Installation view.The exhibition is open till September 15 in Venice

AS: You often evoke probability theory. Are errors the products of calculation?

FI: From statistics and quantitative perspective, generative artificial intelligence produces the average image based on the most probable answer. With error, instead, you get the most improbable answer, the least likely one that can be, in my opinion, the most interesting one, where you can find the most creative scope.

AS: What can these mistakes teach us? How errors influence us on the level of concepts and our perception of aesthetics?

FI: Errors teach us, first of all, to escape from the classical, stereotyped representation. They make us realise how often the new look at something that is at first judged to be a mistake is getting out of the ordinary and becomes a new way of looking at things or showing us the different paths.

AS: Is there still something interesting in the aesthetic of camp and kitsch that can be detected through AI?

FI: There is indeed, if camp or kitsch is an aesthetic part of one’s artistic approach. But only if somebody makes a true creative path out of it, but most mass-generated images are essentially always the same thing.

AS: You are a philosopher by training. But you are an artist, journalist and curator as well. What is your main professional identity?

FI: I have never understood which identity is most important. For me making art means performing a philosophical investigation, it’s a way of doing philosophy with images instead of verbal or written language. It’s a part of the same intellectual journey. Internally, I do one thing even if it looks from the outside as two separate paths. On the other hand, I use images to do philosophy as well. Sometimes I can express a concept or do a philosophical investigation better with images than with text.

AS: You say that every work is an accumulation of experiences, of ideas, of predecessors. Hence the question: what becomes stronger in the case of AI, the probability theory or the free will and the choices of the artist?

FI: Here I have to answer as a philosopher, in the sense that I don't believe in free will and that we are free to choose things. I’m sure that we simply cannot predict our own actions in the future. But even setting aside the physical inevitability of events andother philosophical arguments, the social and political context in which a work of art is created is very important.

The “will” of the artist plays a crucial role as well, without a shadow of a doubt, because it is the one through which the whole social and cultural context is filtered. The final result is unpredictable only because we are not able to foresee it, not because it is given a priori, but every element contributes to it.

AS: We all know that some artists are stronger than others. How does it impact the artworks made with AI technologies?

FI: I believe that every true artist makes a personal contribution that differs from others with respect to the context. Of course it’s not only the context that drives them. Behind every artist lies their own sensibility, their psychology, their life, their history that influences them. Many of those personal things are not chosen, such as native country or language. Certainly the individuality of the artist filters through the whole cultural world and each time does so in a different and original way. So the artist always gives a different imprint to the work.

AS: You say that artificial intelligence accumulates experiences. However those experiences vary, could you develop on that?

FI: AI processes an immense amount of data, which comes from certain parts of the world. Some of it is more Western, some—more Eastern, it depends on where the AI was initially created. Each AI has the limitation of the dataset, which is immense, but it turns out that it was given. So the artist may force this situation and play with this data in such a way as to obtain something unexpected and different. But AI in itself will always be restricted by the data. Humans are restricted as well, we simply possess much more of it.

AS: This experience of predecessors includes biases such as sexism, xenophobia, etc. As you say there are no universal ethics. So who will decide what we should get rid of and what we should preserve?

FI: That's a good question. At the moment, unfortunately, the production companies decide, and they are full of prejudices. They tend to do so by virtue of the era of the place where they are, so an American company will implement American ethical principles, a Chinese company—Chinese ones. The time is also a significant element as now we’re operating ethical criteria of 2024. If we had an AI from the 17th century or from the 1930s, ethics would vary. Moreover we have no idea what ethical criteria we’re going to have in the 2300s.

AS: Is there a possibility to create a truly free AI? Without biases and without prejudices?

FI: In my opinion, no, as it is data-dependent anyway, and all data is always biassed per se. The fact of choosing which data to put in is already a bias. A ‘truly free AI’ is impossible from a logical point of view. Still we can improve the situation. For instance, the more data there is, the more diversification there is in the data, the more possible it is to go in various directions. If instead it gets circumscribed, the more closed the data is and more likely it will end up as a strong bias.

AS: How about the subconsciousness of AI and its difference from the subconsciousness of human beings?

FI: I don't know because we don't really know fully how the human brain works yet. Same goes for AI. Subconsciousness is possible in both cases. Some biases come from a certain accumulation of data in AI, as well as human beings, tend to give the most probable answer. After all, if you ask me, “How are you?” I reply to you “I’m fine”, the most probable answer. In this sense the fact that there is a statistical factor behind even unconscious mechanisms and therefore that the most probable answer is always given, creates a habit that produces a certain behaviour. Then the rest is a bit of a mystery, we don’t know our subconscious, let alone what it does. There's the element that we don't know how it's calculated, so it's still mysterious to us, always.

AS: What advice would you give to artists starting to work with AI? Why are there still so few of them? Is there anything to avoid for an artist starting to work with AI?

FI: Probably the thing to avoid is to stop the process at the very beginning, when you realise that you get immediately the results that do not satisfy you. It’s the first wall to break through. You should understand that AI is not a tool that will provide you a desirable result immediately and easily. As with any other media you have to work hard enough. In short, you must not settle for the first thing. And it’s not as easy as it seems.

AS: What is the potential of artificial intelligence in art now? Is there something undiscovered?

FI: Certainly, the AI potential is enormous. A new tool was born, just as the camera was born in the past. We still have no idea how it will influence art for those who use it and therefore choose to explore this tool.

Moreover it will eventually influence those artists who do not use it. As will still live in a world where artificial intelligence exists. In the same way as it happened with the invention of photography.

AS: AI accelerates quite fast. Does the theory on AI follow its development fast enough?

FI: The theory is much slower for now. And still we are left to deal with the first questions. Whereas technology is moving forward. So we are chasing it, but from a theoretical point of view we are still behind. In my opinion it will take some time.

AS: There is still a lot of scepticism towards AI, including the academic sphere. Does the theory follow the actual AI agenda?

FI: It's more of a practical problem. We still don’t know all the effects of AI and without it we can’t provide decent theory. We can only imagine, speculate, think about what might happen. The theory should be built not only on technologies but on its effects.

AS: And how can we collect the information about these effects?

FI: In my opinion we have to look at those who use them in the first place, the artists. We have to observe the fruits of their work, their reactions to technology and their experiments. If we look at how technologies of the past have changed the social structure we may notice that our reactions are always the same, it's the technologies that are different. However we react to them more or less always in the same way. Someone would be afraid, someone would be passionate and excited.

AS: And what is the role of the institutions including museums and universities? We know that most of them are quite conservative and slow, there's so much bureaucracy. Is there anything we can do?

FI: I would like that, I would be happy. I think we should, in the sense that institutions, especially cultural ones, like museums, universities and so on, should experiment with AI a lot more. If they won’t experiment they won’t be able to study. Many of them stop at conservative positions, without getting their hands dirty, without experimenting a bit. Thus they stay away from the avant-garde of technology.

So I think all institutions, especially cultural ones, should be much more courageous in experimenting with AI to better understand the limits, the problems, the potential, and study it more profoundly.

AS: As a professor what feedback about AI do you get from your students?

FI: I teach at the Academy LABA in Brescia where I show students how to use these tools and how to integrate them with other artistic practices. It is very interesting because in this way you also discover a lot of reactions that you won’t get otherwise. They use AI in so many different ways showing impressive results. In fact it is one of the reasons why I believe they should be taught precisely.

AS: You sound very enthusiastic. Do you feel any kind of scepticism or decay? Looking at what’s happening in the world and how AI may be used. The unprecedented amount of information is also quite difficult to grasp and to make a choice between something precious and ordinary.

FI: Indeed, I’m enthusiastic about artistic use of AI but I have to say that I share many criticisms of this instrument. Obviously there are dangers and very critical points that need to be addressed. For instance the military use is quite worrying, or the pollution and energy consumption. These are quite technical aspects. But there is also the risk that, precisely, prejudices brought with artificial intelligence will spread and will be reinforced in society. And any work-related issues, even if these are not technological but social problems.

The impact of AI on the world is not clear yet, whether positive or negative. They seem to be questions of the existence of technology but in fact they are problems of the society that produces and receives it. And it’s the society that has to react and reflect on it.

AS: What is the real hierarchy between technology and humans?

FI: Actually we can never separate human beings from their technology. If you take language and writing, it’s still a technology, and yet it has also changed us at the level of the brain, precisely the brain structure on the physical level. Therefore human beings can never be clearly distinguished from technology. Maybe we are hybrid creatures?

AS: Are the first results produced by artists with AI less valuable than the later ones?

FI: Generally yes, because at the beginning there is still some learning to do with the machine. So the results of first interactions tend to be worse. Then they get better. It’s like sketches before an oil painting.

AS: Would it become necessary for artists to use AI?

FI: No one is forcing artists to use layout programmes, Photoshop, Midjorney and others. However they have to be part of the basic knowledge, because one day they can serve you. It may happen that you will never use them in life, but they can be useful.

Artists who learn to use AI even if they won’t use it will have an advantage because it is a technology that is helpful. I don’t think that the knowledge of AI replaces everything or that AI steals work from artists. Certainly it makes the field more competitive. AI just should be included in the syllabus of art academies. The more tools you know about, the better.

AS: Would it be truthful to say that thanks to AI the whole structure of the art world became more horizontal, more open to every person, and that everyone now became an artist?

FI: No, I don’t think so. More people now know how to use technologies, but very few use them to their full capacity. It is very similar to photography, that is, anyone can take pictures. It is indeed very easy, but only a few people can take a good picture. It’s the same with AI.

AS: Which skills for artists are most important in your opinion?

FI: Well, in my opinion the most important skills are never technical, but are more mental, that is, creativity is the most important skill for the artist, whether you pour it into painting, into artificial intelligence, 3D, video, drawing. If you don't have a creative mind or you haven't educated your mind to be creative you will always stay behind.

AS: So we can’t learn creativity from scratch, can we?

FI: From scratch or not, you can develop it. Like muscles. There are people who are born stronger, the others may train. Musicians require constant practice of their skills as well.

Read more:

La rivoluzione algoritmica delle immagini(The algorithmic revolution of images Art and artificial intelligence)

Francesco D'Isa, published by Luca Sossella Editions, 2024.

https://lucasossellaeditore.it/libro/arte-intelligenza-artificiale/