Source: Collection Weekly of Guangdong New Express Daily   Published: October 16, 2022

The marriage of art and technology has always been a heated topic. Although the emergence of photography in the 19^th century had an impact on the style of painting, it does not seem to be as profound as this time when A.I. intervenes in artistic creation. With the news of an A.I.-generated painting winning an award in the U.S., the internet has once again seen a climax in the discussion of A.I.-generated art. But whether A.I.-generated painting can be considered as art is also a point of controversy in the industry. Thus, can we analyze the impact of A.I. on painting from the relationship between human and art at the very beginning of art, as well as the changes of such relationship in the development of art history? Feng Yuan, professor of the School of Communication and Design of Sun Yat-sen University, said in an interview with Collection Weekly that artistic creation is not mathematical logic, so it cannot be replaced by A.I..

Jason Allen’s A.I.-generated work, “Théâtre D’opéra Spatial,” took first place in the digital category at the Colorado State Fair, making it one of the first A.I.-generated pieces to win such a prize, and setting off a fierce backlash from artists who accused him of, essentially, cheating

Mr. Allen created his artwork with Midjourney, an artificial intelligence program that turns lines of text into hyper-realistic graphics. The New York Times reported on this in September, “A.I.-generated art has been around for years. But tools released this year – with names like DALL-E 2, Midjourney and Stable Diffusion – have made it possible for rank amateurs to create complex, abstract or photorealistic works simply by typing a few words into a text box… Mr. Allen declined to share the exact text prompt he had submitted to Midjourney to create ‘Théâtre D’opéra Spatial.’ But he said the French translation – ‘Space Opera Theater’ – provided a clue.”

Collection Weekly: A.I.-gennerated painting has been a hot topic again recently. Is it of interest to you?

Feng Yuan: The performance of artificial intelligence in painting has been particularly striking in recent years. I think we can divide A.I. into two major types: one is represented by AlphaGo, and it defeated the Go champion, which shows that in terms of algorithm-based intelligence, A.I. is bound to surpass humans; another kind of A.I. is represented by painting and artistic creation, it does not have the absolute nature of mathematical logic like Go, and it relies on human perception and has a high degree of ambiguity. Here, it's hard to say whether A.I. will beat humans, and it's certainly been the subject of much debate.

Collection Weekly: Some art history researchers used to think that “art history is the history of the development of tools”. What do you think is the relationship between art creation and tools?

Feng Yuan: I think we should say that art history is probably the history of the interaction between the mind and the tools.

The main reason why human beings create art is to express the mind; our mind needs sensory expression, and this is the basic motivation for the creation of art in the first place. The most difficult problem in human communication lies in the communication between two minds; because the human minds cannot communicate directly, various “mind transmission media” must be invented, which is the driving force for the birth of painting, and also reveals an eternal problem, which I call “the problem of knowing the mind”. This is what our ancestors meant when they said, “In drawing a tiger, you show its skin, but not its bones; in knowing a man, you may know his face, but not his heart”.

Collection Weekly: So, art is often a medium of communication between minds. In terms of the relationship between minds, art is a real signal sent from one human mind to another, right?

Feng Yuan: Converting the mind, which is not directly observable, into a medium that our senses can perceive, and thus making it a message that everyone can receive – this is art. It has two characteristics: the first is the inner mind, which we call mental ability; the second is the outer medium, which has capacities of materialization. So, on the one hand, art is an expression of the inner mind, and on the other hand, once it is formed into an artwork, it becomes a tool and a medium. As a result, we can see all art as an external medium of the inner mind.

In this regard, I think artificial intelligence will also face a “mind problem”, because the human mind has emotional characteristics. The reason for human emotions is the anatomical evolution of a brain tissue called the “limbic system”, which is possessed by mammals or primates like humans. Only with the limbic system does the brain perceive emotions.

In addition, human perception of emotions has an endocrine basis, and love and hate are endocrine-related. Therefore, it is often said that artists have some kind of passion or some special, almost crazy state, which means that artistic creation cannot be separated from the endocrine-driven neurochemical mechanism.

Collection Weekly: This kind of fickle mood change seems difficult for algorithms to reach.

Feng Yuan: Yes, art creation is not done with algorithms. It's not the same as mathematical logic, and an algorithm based on mathematics may not require emotion. Having said that, it is important to ask a very serious question: if we can create an A.I. that also has emotional changes and endocrine mechanisms, it can be believed that only then the A.I. has a human heart, then of course it can create art. Otherwise, the image processed by algorithm can only imitate but not have the function of creating and perceiving art on its own because it does not have a human heart.

Collection Weekly: It seems difficult to tell the difference between art created by A.I. and art created by humans.

Feng Yuan: The level of art is a subjective judgment, which is based on the type relationship between culture and emotion. In the human world, different cultures have different and unique interpretations of art, but it can be admitted that humans do have commonality and innate nature, which is the biological basis for art to cross borders and cultures, so that people from different cultures can appreciate it. I believe that the current and future A.I. is certainly able to imitate human painting, and even surpass human painting in terms of technical difficulty, but if A.I. has no emotion, then it will always lose to humans in terms of emotional expression, and whether painting as a medium moves people, after all, is a mystery that only the human mind can master.

So, in the sense of medium, A.I. can draw a lot of paintings, but most of them are imitations of human creation. Even if A.I. can create autonomously, but why should A.I. create it when it is not driven by emotions? This is not a question of the technical level of the medium and tools, but in terms of art being an interactive creation between the mind and the medium, unless an “artificial heart” is designed for A.I., it will never be able to experience and understand art as humans do.

Collection Weekly: Thinking back to the 19^th century when photography emerged, was it perhaps also a challenging technical innovation for classical painters at that time, which led to the emergence of Impressionism as well as Modernism? If we look at it from this perspective, does A.I.-generated painting have a counteractive push on art creation, just like photography did back then?

Feng Yuan: This is indeed a question worth thinking about. The invention of photography and its popularity have indeed changed the development of painting significantly. For the past tens of thousands of years, the visual pursuit of human beings has been aimed at reproducing the world. The emergence of photography has provided a better solution for the pursuit of reproduction and duplication, and at this juncture of technological alternation, it has indeed prompted a subversive turn in the visual arts that is mainly based on painting.

It is important to know that the earliest human paintings that can be identified today date back to the Ice Age, about 30,000 years ago, when the steppe bison murals painted by human ancestors inside caves were figurative paintings. Despite the limitations of materials and technology at the time, it did not affect their ability to draw vivid outlines and images of the steppe bison, which shows that the visual medium created by humans from the beginning was figurative. Why? Because human visual thinking and artistic pursuits come from the survival benefits gained from evolution. Obviously, only painting figurative paintings was conducive to the survival of primitive people who depended on hunting and gathering for their livelihood, and if they had painted abstract paintings tens of thousands of years ago, they might have become extinct long ago.

Replica of the painting from the Chauvet cave. The original art is approximately 31,000 years old, probably Aurignacien

So, conversely, why did abstract painting come about as late as after the 20^th century? It is not that abstract painting is more difficult to paint than figurative painting, on the contrary, only when we have crossed the threshold of survival, when human beings have realized the benefits obtained through figurative media, that is to say, to the modern society after industrialization, abstract painting, like a luxury item, came out of the blue and greatly enriched the diversity of human visual perception. Therefore, the invention of photography did not abolish painting, but freed human artistic creation from the existential benefits of figuration, prompting humans to re-establish the symmetry of ambiguous cognition and emotional expression. In today's era of A.I. advent, I think the inherent truth is still the same. A.I.-generated painting will not put an end to human creation, but will promote human beings to seek the relationship between mind and medium again, to seek known or unknown ways, or to give full play to the imagination of human mind and expand the diversity of mediums for human emotional perception by transcending the characteristics of figurative-abstract medium.

Collection Weekly: This explains the question why abstract painting was never mainstream before photography.

Feng Yuan: Yes, it wasn't that humans couldn't paint abstract paintings at that time; it was that humans at that time wouldn't give any value to abstract paintings, because abstract paintings couldn't help humans to survive. And it is worth noting that advanced technology will also in turn imitate the habit of backward technology. For example, later on the early development of photography, for quite a long time, in order to make people accept the artistic expression of photography, photography actually imitated painting in artistic expression – this phenomenon is also very interesting. It expresses the fact that human cultural judgment is emotional and patterned, so not only did the new technology not immediately change the path of the emotion-transforming medium, but in turn the new technology had to imitate the old ways. Therefore, we can speculate that although A.I. is a very remarkable new technology, but because it has no emotional mechanism itself, when the programmer prepares its creative program, it will inevitably imitate the pattern of human creation driven by emotion, and that the cleverest algorithm cannot replace humans to experience the good and bad of art in uncertainty and ambiguity, or to be moved by art. In a word, A.I. does not shed tears, so it cannot be moved, and therefore it cannot have the perspective of the mind.

A.I.-generated painting “Old Beijing Venice” is modelled on the work of Canaletto, the Italian painter from the Republic of Venice (A.I. tool: Stable Diffusion, design assistant: Simon)

Collection Weekly: Photography has not eliminated painting; it has only changed the direction of its development. What will happen with the intervention of A.I.? Can you tell us more about the future direction?

Feng Yuan: First of all, A.I.-generated painting will increasingly reach and surpass human level in terms of media technology, but I am afraid that it will never be able to completely eliminate human painting, because A.I. cannot perceive art autonomously and cannot have emotions like humans. Secondly, we must also admit that with the intervention of A.I., painting will definitely usher in a second transformation, because, in terms of algorithmic superiority, A.I. is able to do at any time what human beings can only do after spending a lot of time and accumulating a lot of work; for example, if you want to paint in any style or in a certain symbolic form, algorithm can easily achieve it. Combining these two points, the conclusion comes out that, firstly, A.I. is able to perform and complete all the art styles that humans can create, even surpassing humans in terms of the difficulty of media and materials; secondly, if A.I. does not have a perceptual mechanism similar to the human mind, then it is difficult for it to create styles autonomously, even if it can do so completely from the technology of manipulating materials and media by algorithm. Thus, before the birth of the “artificial mind”, even the most advanced algorithms cannot make A.I., which lacks a mind, surpass humans in artistic perception, and the style created by A.I. still has to be modeled on the human mind, and ultimately has to be evaluated by the human mind.

Let's look at photography again. Technically speaking, photography can be done without the model of human mind, because photography is based on the principle of light sensitivity in the chemical sense of recording light as an image, and because of this feature, it can be said that photography surpasses human painting in reproducing reality, and it can record the light of the external world more intuitively and accurately. In contrast, A.I. is obviously not as simple as photorealism, but whether it can surpass humans in terms of value also lies in whether A.I. can have autonomous value judgment. As I have already said, artistic judgment is a specialty of the human mind driven by emotions. The biggest problem of A.I. is that it cannot abandon its imitation of humans and create a new style autonomously; even if it can create a new medium, it must be incorporated into the feelings of the human mind before it can be recognized as art. It is still difficult to make an accurate prediction as to which direction it will go.

A.I.-generated paintings are extremely popular with cyberpunk fans: “Kingdom of Heaven” by Henar Sherif

Collection Weekly: What kind of attitude do you think artists should take towards A.I.-generated painting?

Feng Yuan: For today's A.I. technology, using A.I. for art creation, we can only see it as a change of media technology, which does play a great role in enriching art forms. In terms of external media, it can even bring a whole new way of production, for example, on the basis of advanced algorithms, just by entering a few simple keywords, it can create a variety of styles of paintings, which can be imitated in a few seconds, and can even match the real thing.

However, when we look at the logic of A.I. clearly, there is no need to overstate its significance. Artists can safely say that although A.I. can draw something more like paintings than those drawn by humans, whether it is artistic or not, whether it is touching or not, that is still up to you and me to say.