The AI Wars II - Evolution of Creative Goods
To uncover what lies behind the current sentiment surrounding AI art, understanding the how's and why’s is essential. In the second part of The AI Wars, we take a deeper dive into the technology of AI and the arguments often used against it.
💡 This article is a part of a two-part series. Catch part one "The AI Wars I - New Tools, Old Fears" here for the full picture.
Let’s go deeper down the AI rabbit hole. How can machines all of the sudden put out seemingly creative results? We need to have a quick look at the principles of this technology.
Deep learning is a type of machine learning that uses a special kind of neural network, called an artificial neural network (ANN). In the case of deep learning, these ANNs are extremely deep and complex. The more layers there are in an ANN, the better it can extract features from data and make decisions based on those features.
The name "deep" comes from the idea that you have multiple layers in your artificial neural network—and each layer contains many neurons (or nodes). The first successful use of deep learning was Geoffrey Hinton's backpropagation algorithm for training multi-layer perceptrons (MLPs).
Since then, researchers have expanded upon this method to create more advanced networks with higher levels of accuracy in image recognition tasks such as ImageNet classification, which involves identifying objects among thousands or millions of images and speech recognition tasks related to spoken language understanding.
So in general, whenever we talk about AI, it is far from an actual intelligence. The machines we use are not self conscious. They are learning machines with amazing capabilities - but in the end they always rely on both a huge database and a set of instructions. The reason we often get the impression that there is something like an intelligence at work is that we can not even start to imagine using such a huge base of data to reach decisions. It is simply beyond the scope of our minds.
A Very Certain Randomness
In fact the big breakthrough in this field of science is that there isn’t a valid single answer in such a system. It is more like an educated approximation. And it allows multiple answers to a single question. This is why, when using a text to image machine, the same input will yield a different output each time you use it.
The breakthrough in AI relies on adding a certain randomness to the equation, and quite literally so. The ingredients to an AI generated image are not only a huge database, a fixed set of instructions, but also some instructions that are randomly filled with a value out of a defined range. Some people compare this fact to human creativity, but while it often results in a similar outcome, the cause of an effect is completely different.
Humans, especially those amongst us that we consider artists, have strange thoughts - and even stranger ways to produce these thoughts. Still, whenever we start to be creatively active we make an infinite number of minute decisions. We are not consciously aware of them. We just feel that it is right to put that brush onto the canvas on exactly that point and move it in this specific direction. A part of it is learned skill, a part is experience gained over the years.
We describe the process with words like “intuition”, “a feeling”, maybe “instinct”. All these concepts are still not computable. We don’t even understand the processes ourselves, let alone are able to put them into computer code.
So Are the Critics Right?
So, if machines can only rely on what we feed them, can they - at least right now - make art at all? Did I just agree with all the groups of AI art critics? Let me answer like this: Yes - and no!
Yes, computers can make art - in the same way a brush can make art. As a tool in the hands of an artist a machine can do wonderful things. And if you are now saying: „Then it’s the human, not the AI doing the artistic process“, I couldn't agree more.
This is the argument I would like to use against the critics in our Group B from before. AI output can certainly be art. And in current systems there is no output without human input. So inherently there is no creation by an AI system without human interaction. So we move right on to Group C before we tackle the bigger controversy of them all.
Can AI Art Have a Value?
So is AI based artistry just a novel entertainment - or does it contribute to our collective treasure of art? This is something that has a lot to do with the eye of the beholder, and again something that has been at the base of controversy many times before in human history.
The reception of art is deeply rooted in culture and a result of Zeitgeist. One (wo)man’s trash can be another (wo)man’s treasure. It depends a lot on where and when your cultural roots lie. Art is subject to definition. And it is constantly evolving. What we consider art today would in many cases not even have been recognized as art a couple of hundred years ago. We can not anticipate the view of a future humanity when it looks at our humble creations.
Art - especially modern art - is judged by two groups. The first one is a group we might call the experts: Professors, gallerists, curators, you know what I mean, people in turtlenecks sipping on champagne at your next vernissage. They all have a voice - after all they spent a lot of time with and around art and have an opinion about it.
But then there is the other group: The art buyer. The average person buying a piece of art for one sole reason: They like it.
Fortunately not all collectors see art as an investment. They simply fall in love with a piece - or an artist. And the number of these new collectors is on the rise, which is largely a result of the liberation of digital art brought about by the rise of NFTs.
NFTs make digital art a uniquely identifiable asset. This made it possible for digital artists to sell their art as a single piece - or a limited edition of pieces. Now collectors pay for something that is still not understood by many.
Digital Goods for Digital Nomads
So why should I pay for something I could simply copy? The old argument of a generation used to „Google-shop“ the next background for their desktop or their PFP for the next social media platform.
The reason is simple: Because of appreciation for the artists and their creations. Just to not steal from them.
Today’s new generation is also more digital than the older generations would have ever believed possible. For the digital nomad of today, who likes to travel light in this life, any physical item is a hindrance. Why not keep what you love in the digital realm, be it your tablet, laptop or cloud?
Concepts of life and possession collide hard these days in between the generations, but one thing is certainly true: Just like digital art does in general, AI generated art can have a value. How so?
The buyers decide what to spend their money on, and very rightly so. It’s like the old trading game you play as a kid. You give me this - and I give you that. As long as both parties agree, there is nothing wrong with this.
I do agree with the critics in Group C on one point though. That is that the tools used should always be communicated clearly. This is something you would want in the real world art market as well. After all, when you expect „oil on Canvas“ and get „printed on paper“ - you very rightfully feel deceived.
So it is up to artists using AI to properly communicate and mark their work for what it is. And the possibilities are endless… we are long past the point where there was only „AI made“ and „hand made - or non AI made“. Artists have developed beautiful ways to use AI only for certain stages of creation.
So it is time to tackle Group A’s killer argument:
Using AI is Theft and Plagiarism
From „you are a fraud“ to „you kill poor artists“ - most of us using AI in some way or another have had to read harsh accusations in the socials. It’s about time we dive into the big question: Is using AI really theft and plagiarism?
What a deeply philosophical question this actually is, is not even recognized by a lot of people spreading hate amongst artists on social media. True, when you look at the database used by your art AI, it is full of other people’s work. After all, nothing comes from nothing. Including art. But is this the end of the discussion - or just the start of it?
As you might have guessed, in this case it is the start, and I want to share a point of view I learned from a conversation with an art professor. This talk was about how a human art student learns basics in order to become an artist.
The skills required are - amongst others:
A - Know Your History
To be able to go forward, you should take a close look at those who came before you. By studying old styles, materials and techniques you understand where certain artistic expressions comes from
B - Look Beyond Your Own Culture
Look at the arts as a whole. Study different expressions of creativity based on diverse cultural backgrounds.
C - Copy and Learn
When learning the basics you will go through training that has largely been in use for a long time. Many students did it before you, many will after you. Even if style, material and techniques change, the basics are still the same. Composition and proportions are still the same, no matter if you use a stylus on a tablet or a pencil on paper.
Once you have reached a certain basic skill set, you will be encouraged to not only go out to museums to look at the work of those considered masters in their field, but to outright copy works, just to practice!
So, let’s compare…
AI vs. Art Student
At first glance they don’t seem to have that much in common. But in the second place they are very similar.Deep learning and studying are quite comparable. If you look at what the AI does, it has everything in common with the human student. Yes, art Als can and do not create something completely new and groundbreaking out of thin air, BUT neither can a human!
Before the typical human artist starts an artistic career, she or he has been thoroughly influenced their entire life. We process millions of pieces of sensory input on a daily basis, so we are always in part a product of our environment and culture. The effect becomes even more profound when we decide to actually study the art of observation and creation that is called art.
If we like it or not, we are all standing on the shoulders of giants.
So now it is time to take on Group A, the group that is claiming using AI is theft and plagiarism.
After considering all that has been said I firmly oppose this as long as it is generalized. AI can very well be an immensely powerful tool for an artist to express their thoughts and emotions. And there are so many possibilities to make good use of it. But I must also admit that there is a dark side, a variety of options that are indeed hurtful up to being downright criminal.
The balance tips from neutral to bad the moment you decide to downright copy - especially if you do not communicate properly. As everything in life there is no black or white but a ton of shades of grey.
An example: We all know pictures that copy a certain painter’s style that is being transferred to another picture. This is a practice even done way before the rise of digital art or AI.
Most of the time these are works of painters long gone that are in the public domain right now. As long as you don’t claim your art to be an original by that specific painter, you don’t hurt anybody. Basically it is generally seen as being an homage, a celebration of the well known old art.
But things look very different if you are using styles of contemporary artists, especially those working fully digital. These individuals have worked hard to create a recognizable style which is their base for earning a living. These artists are very rightfully shocked when seeing their personal style of art easily copied by using AI tech, so we do have to develop an etiquette for the use of AI technology.
Artists working with these tools should be very aware of their input into the AI programs. Best practice is to never use the name of an artist in describing what they want the result to be.
Sure, it takes a lot of time and research to find out how to best describe what you would like to see as a result. But this journey is a very rewarding one. It takes you deep into the world of art itself. You will learn to differentiate between styles and what their specific traits are. You will certainly get to know more about artists you never knew. All in all you will become a better and more complete artist.
Conclusion: Use AI Consciously And In a Fair Manner
I can only urge all artists to be fair to one another and to your collectors. It is fine to use these emerging technologies. But be transparent about what you do. Don’t claim to be able to do things you really can’t. This will always backfire in the long run.
AI is here to stay, so learning about its capabilities and keeping track of its progress is a wise course of action. I am confident that developers of AI systems will find better ways to exclude certain artists’ work from their databases. But nonetheless, the responsibility always lies with the artist in the end.
It is up to you to make the best and fairest use of this technology.
Keep creating and find your own artistic expression!
This article is written by Luvcraft as a part of enter.blog's bounty program. Do you have an interesting topic, series or subject you think would be fitting for enter.blog?
You can now submit your articles to enter.blog and get paid for your contributions!
Read more and submit your article here.
Looking to learn more about NFTs? In this section you can learn more about how NFTs are revolutionizing the world.