As a follow up on my previous article, “The importance of fine art photography”, I wanted to share my initial thoughts on AI and how it can impact, be it negatively or positively, photography and art in general. By all means, this will obviously not be my definitive thoughts on AI and art and I might change my thoughts the same way AI will develop and integrate into our lives. I hope to share more thoughts in subsequent articles.
To start with, let me state clearly that I’m very excited to encounter and live any change in our lives and the world we live in. Of course, positive change is always preferred to negative change but it doesn’t take away of the fact that change means life. The only things not changing are things that are not alive, that are dead. Change also often means progress. Not always though, but most of the time, yes, change is positive although we may not always recognize it as such initially, when confronted with big changes. Especially change and progress in technology is welcomed at all times as it’s tightly linked to progress in civilization and humanity.
Therefore I’m truly excited to see the recent major breakthroughs in AI and how it affects all parts of our society. And I think we should all embrace it, without a doubt. This initial thought I have about it, will most likely not change. Not in the near future and not on the long term. Change, after all, is the essence of life. Even if some of the unintended side-effects seem to be negative.
AI image generators
I am not afraid of AI. I am not afraid that it will replace me as a fine art photographer. I am not afraid that AI can soon create images that are as good as the best images photographers can create and might render me redundant. In fact, there are already images created by Midjourney and the likes that are simply mind blowing and often look better than what the majority can create. ‘Look’ in bold and in italics, because I am only referring to the looks of it, not to the intrinsic value of it, and I get back to this important distinction further in this article. I am only afraid how people will use and abuse AI to manipulate us and even set us up against each other.
Prompt used to generate the image: black and white fine art photography, architecture, 3-d rendered buildings, higher contrast around building, apocalyptic background, soft ethereal background, art-deco style architecture, dark mood, fritz lang poster style
Of course, there are several issues that arise with the advent and development of AI, from a legal, intellectual rights point of view, from a societal point of view, a political point of view and so much more. We need to live with that and adapt to the new situation, and we will. And also in art, there are several fundamental questions that arise, but also here I believe we need to think critically and see how to live with it, how to embrace it and how to use it.
So, what do my initial thoughts entail more than what I already stated?
What is AI
First, let me give a short description of what AI is and then try to answer the question if AI can create art. And to make this even more topical: I used Bing Chat to give me a short description of what AI is. This is what it answered:
[…] Artificial intelligence (AI) is a term that refers to the ability of machines or software to perform tasks that normally require human intelligence, such as recognizing faces, understanding speech, or making decisions. AI is also a powerful tool for creating images, sometimes called AI art or promptography, that can mimic or surpass the quality and realism of conventional photography.[…]
That’s a useful description of what AI practically is. What it intrinsically is, that is another more profound question but for now let’s stick with Bing Chat’s description. And I already want to note what I’m doing here when using AI: I’m specifically disclosing that I have used AI to give you the exact quote and for a reason: when using AI and you use the generated content, you should cite your sources. When you create images that are not photos but AI generated images, you should disclose it and not let people assume they are photos. I don’t pass any judgment whether you create images with AI or with a camera. But you have to disclose it especially if there can be any doubt.
Can AI Create Art
Now, can AI create art? It depends on the definition you give to it, but according to my definition: No. It’s my strong opinion AI, as it is now and I believe for a considerable time to come, and as long as it doesn’t have a consciousness, cannot create art. It can create beautiful images, but not photos (at least for now), let alone art.
This demands an explanation, even though my thoughts on this aren’t fully formed yet. But I will explain why and what I strongly believe at this moment about AI and art. And photography.
First of all we need to define what art is. Instead of asking ChatGPT what it is, I’ll give you my thoughts on what I think art is, based on what I’ve read and learned over the years and then add my own personal experience and preferences to it. Essentially, it is almost the same learning process as what AI does: learning from available sources. But here we hit an essential difference: I’m adding my own personal experiences, emotions, and preferences to it. Which are elements that AI cannot add to it. More on that later.
My Personal Definition Of Art
A definition of what art is, my personal definition:
After writing this article, I asked ChatGPT what art is and interesting to see that it contains many elements that are also in my definition of art. Also interesting it emphasizes ‘human skill’.
I am of the strong opinion that one very convincing way of proving an entity has consciousness, is through an expression of art. Hence, wherever art is being created, we know there is consciousness. An experience can be anything from physical pain and joy to mental excitement.
Why do we create art? I’ve tried to answer this question in this article but for this article let’s stick to the topic at hand. My definition of course is not an absolute definition but it does well for me for now as it covers everything I believe art should be.
Let’s break down the most important elements of the definition.
- ‘Self-initiated expression’. We, as humans, we create art because we need to, we feel like it, because we have this urge to express ourselves and not because someone asks us to create art. It comes out of ourselves, from within not from outside. If AI could create art, it surely doesn’t come from inside AI, it creates ‘art’ because we ask it to. We give it a prompt. Apart from that, it doesn’t feel a need, an urgency to create art. It feels nothing until we ask it to ‘feel’.
- ‘Autonomous conscious being’. Our human autonomy, no matter how imprisoned we are, physically or mentally, ensures we do what we want, when we want it. Ensures what we express when we want to express. It is because we are conscious of who we are and that we experience this world and this life, uniquely through our (mind’s) eyes only. AI doesn’t express what it wants to express because it doesn’t want anything. It only executes when we ask it to. AI doesn’t spontaneously do something. It doesn’t feel the need to, because it doesn’t feel. It doesn’t have a consciousness.
- ‘Deliberate intention to communicate an authentic and individually lived experience’. We create art by creating on purpose. It is no happy accident. We don’t do this subconsciously or unconsciously even though we are not always aware we create art. We create something consciously. Because we can. Because we feel the need to show we have something internally as a result of a subjectively lived experience that needs external recognition. AI only creates ‘art’ if we ask it to. It doesn’t create it spontaneously, because it doesn’t feel the need to show it has an internally lived experience, because it has none.
- ‘Stylized and aesthetic way’. When creating art we want to create it in a way that suits our aesthetic preferences and ideals. We want to give shape to it. When AI creates art, it does that too, but not because it has an aesthetic preference. It creates something in ‘the style of something’ without a preference.
- ‘Has the potential to move and inform the observer’. Usually this is the case when we, as an observer, experience something through art, that we haven’t experienced before. When we experience something through art we experienced before then we are experiencing a cliche. This goes deeper than just the visible surface. It is also about the feeling that is evoked and we experience as an observer. The artist feels something and tries to express that. Sometimes the feeling comes across, sometimes it doesn’t. AI doesn’t feel and can’t express that. AI generated images are at best aesthetic but lack substance. They lack personality.
This is why I believe AI can never create art.
Consciousness and its relation to the physical
To elaborate a bit more on the phrase ‘individually lived experience’: it is important to note that we live an experience, or experience a life, that we are conscious of in varying degrees, not only mentally, but also physically. Sometimes deeply aware, sometimes only superficially aware. We see through our eyes, we smell with our olfactory system, we hear with our ears, we taste with our taste buds, and we feel the tactility of objects through our touch receptors. When we pet a cat, we know what it feels like. We know how it feels when we talk about furry. AI can be trained to know what furry means, but it doesn’t know how it feels. When someone pinches our arm, we feel physical pain. When I hit my monitor with a hammer, it won’t feel anything. When someone breaks our hearts, it feels very real even though nothing physical has been broken. When nothing physical is broken in AI, nothing at all is broken. AI doesn’t have a tongue, hands, feet, arms, or legs. It doesn’t have a heart. Consciousness is not just what’s constructed in the brain. Our physical qualities and abilities form an important part of the construction of consciousness. When I talk about lived experience, about consciousness, then you can’t separate the physical experience from the mental experience, they both play a part in the forming of a highly individual and one-of-a-kind consciousness. AI doesn’t have all that, yet. And I don’t think it will ever have.
Therefore, AI cannot create art. It creates artistic-looking results, but it doesn’t create art. It creates something empty, something that doesn’t come from a truly lived experience, something that AI doesn’t even know itself if it is beautiful or not. It creates artistic-looking results without any opinion about those results. It is just the most likely result given the parameters and the prompt. We create artistic-looking results because we have an opinion about what we like, without a prompt.
Again, after writing this article I asked ChatGPT why it included the phrase ‘human skills’:
Differences in processing information
Related to that, there is something essentially different in how we humans process information compared to AI, something Noam Chomsky in a recent essay in the New York Times highlighted: […] The human mind is not, like ChatGPT and its ilk, a lumbering statistical engine for pattern matching, gorging on hundreds of terabytes of data and extrapolating the most likely conversational response or most probable answer to a scientific question. On the contrary, the human mind is a surprisingly efficient and even elegant system that operates with small amounts of information; it seeks not to infer brute correlations among data points but to create explanations. […]
Humans want to explain information, AI doesn’t, it just uses it. When Picasso studied Matisse’s work, he asked himself why did he do it like that, why not like this? When we study Ansel Adams’s work we ask ourselves why like that, why there, why not like this and here? We do not just want to emulate but we also seek explanations, what it is to emulate in the first place, and with the explanation we give ourselves we are able to create something new. The need for explanation is a difference that enables us to create unique and authentic art.
Humans make infinite use of finite information, AI needs infinite information to make use of it in a finite way.
More ChatGPT chats:
Embrace AI as it can give birth to other art forms
AI is a tool, a medium and I treat and embrace it as such. It can help us create art, it can help us find more ways of expressing ourselves and exploring ourselves and the world we live in. I’m actively exploring how AI can help me in what I create and everything else AI may be able to help me to get the most value out of the life I live. But AI in and by itself cannot create art.
Therefore, like I stated before, I am not afraid of AI, I am only afraid of what people might do with AI. I’m not afraid AI may replace me. If anything, I believe art will always be one of the few sacred places AI will not be able to replace humans.
So let AI come into our lives and I will try to make the best use of what I can do with AI and create and maybe, or better yet, certainly, AI will give way to new art forms in which humans will use AI as a trigger or a tool to explore and develop those new artforms. Just like painting transformed from the literal to the figurative, from the realistic to the abstract forms of painting, after the advent of photography. Photography never replaced painting, it just gave way to new forms of painting. AI will never replace photography, but it will give way to new ways of creating images. But be aware of anyone who uses AI to manipulate us. In politics, in news media, but also in art.
AI learning and intellectual copyrights
On a related note, but since it is not the essence of this article, I will treat it as a side note, I want to address intellectual copyrights and AI.
There’s increasingly strong opposition against how AI is being ‘fed’ images from the Internet and other digital sources to train it without the consent of the original author. I’m not exactly in the know how the images are being used to train AI and there might be a legal issue there depending on how those images are being used but let’s just assume that AI just ‘crawls’ the Internet and memorizes these images. Just like we humans essentially do: when we create art through images, then it doesn’t come out of nothing. All artists have been standing on the shoulders of those before them or have been inspired by their contemporaries. From Picasso to Van Gogh to Da Vinci and Michelangelo. They have all used some ideas, concepts, and elements from others and put them together in their own personal way to come to something unique and original. Creating something unique and original doesn’t mean creating something out of nothing, without knowing anything of the world around them. It means knowing how to deconstruct existing ideas, images, and concepts into smaller elements, then questioning them and finding a way to reconstruct those smaller elements into something new as a whole by giving ourselves explanations. That is creation. That is how all art is created and how it evolved.
So we humans basically follow the same learning process as AI: we actively find information, we actively learn from it and we actively find ways to memorize it and use it in our own work. I’ve seen thousands if not millions of images in books, movies, magazines, and later in digital sources and on the Internet that consciously and subconsciously formed my artistic preferences. When I create my architectural images, it is a result of that. The result of years of ‘training’ and feeding my mind with images. My contribution, or talent if you want to call it like that, is that I have tried to deconstruct what I’ve seen and experienced all my life and then tried to reconstruct all those smaller elements into something new and mine.
AI does the same. What Dall-E or Midjourney create when they’re given a prompt, comes from combining all those images they have seen into something new.
It is argued now that AI shouldn’t do that. Why not? We do the same when we are creating. It is a weak argument and I think we should just let go of that argument to use against how AI is trained and how it comes to its visual results. It denies how we as humans create art ourselves. There might be other legal issues, maybe AI is trained using images that aren’t publicly available and have been illegally obtained, or images have been used in a way that is not permitted. That is another potential issue, but not the point of this. The point is how AI is trained and creates images. And it is in the same way as we humans are trained and how we create images. But we humans do it better. We have our lived experience to give it more value. We can create and think outside the box by questioning what came before and giving ourselves explanations.
And last but not least: we don’t need a prompt to start creating images. We just do it when we feel like it. AI does not.