What Is the Art of Artificial Intelligence?

How a New Impressive Tool May Revolutionize Art Even More Than Photography Did

Sandro Mikautadze
Predict
10 min readJul 23, 2022

--

A robotic human painting in an art gallery

By definition, art is in constant evolution. Movements and ideologies come and go. But, at times, some changes go down in history. Now it can be one of those cases.

Joseph Nicephore Niepce, a French researcher, took the world’s first picture ever in 1826. Little did he know that photography would change art — and the world — forever. Yet, he felt that it was “the first uncertain step in a completely new direction”. He could somehow smell an incoming revolution.

Two centuries later, art is about to witness again a new far-reaching evolution, possibly never experienced before.

Not because of photography, neither due to art itself nor due to artists. Art will change because of artificial intelligence (AI).

If you don’t yet smell the incoming revolution, don’t worry. Let’s try to uncover together why and how AI will transform the world (of art), perhaps even more than photography did!

A view from a window, black and white
Niepce, “View from the Window at Le Gras”, 1826. The world’s first picture.

Why should AI and art have a link?

I know you asked yourself that question. And rightfully so. After all:

  • AI performs only specific tasks that can be mathematically identified; art, instead, deals with creativity, which is all but mathematical.
  • AI is literally 10 years old; art, instead, has existed since the dawn of the human age.
  • AI is full of boring nerds wearing black hoodies; art, instead, is full of happy creative, and inventive people.

Just kidding. The last point is stereotype. Don’t ever think again that I’m about to become a boring nerd wearing black hoodies!

Anyway, the gist is the following: these two worlds are too far apart from each other.

But what if I told you that it is not true?

I know it is a very bold statement, literally. But whether you believe me or not, I kindly ask you to do one thing: go to the top of the article and look at the cover image — the very first one. Look at it and then come back here.

Have you seen it?

Are you sure?

Ok. I trust you.

Go on reading.

That image was generated by artificial intelligence!

Yes, AI did that. It is a sort of android painting in an art gallery. No human being has ever spent a second of his/her life creating that piece of art. But an AI algorithm did.

In other words:

AI did Art.

Do you still see these two worlds too far apart from each other? I hope that by now you will agree with me that AI and art seem to have a link. If this is still not enough…

…let’s try to dig into it.

That image was created using an image generator. The most famous of these is DALL·E 2 by OpenAI and it has been released a few months ago. Others have become available just recently. They include MidJourney — used for the cover image — Craiyon and others. There is even Google, but its project is not public yet (rumor has it that it is the best one).

How do they work?

If you’re eager to discover what is behind image generators, I’m sorry to tell you that I won’t discuss it in much detail. There’s plenty of information online that explains everything much better than I would.

Instead, I’ll try to be as simple as possible.

What do you have to do?

You, a human being with almost non-existing artistic capabilities, have to do the following things, in order:

  • Open the image generator
  • Input a text prompt
  • Wait a few seconds
  • Look at the generated images

Nothing more.

If you were worried about doing some strange math or writing code to get the images, I hope your fear is now gone.

It is all just like searching for an image on Google. The only significant difference is the following: you do not search for images, you create them.

It is as simple as texting a friend and reading the response — actually, this is not true. Some people take days to respond. So, AI creates images faster than people answer text messages.

What does AI do?

Those “waiting seconds” represent the true culmination of state-of-the-art techniques in AI, developed over the years. But they are not easy to explain for different reasons.

Firstly, it’s not easy stuff. You need to know many things about AI to deeply understand what is going on.

Secondly, each company might utilize different techniques. For example, this is how DALL·E 2 — the best image generator so far — works in simple words: the images are developed using a neural network model called a transformer, trained mapping images to text and vice versa. The neural network uses a technique called diffusion. Here random pixels are created and modified step by step, in order to match the text prompt as best as possible. This is how you get your image. But Google’s Imagen uses a slightly modified approach and other firms might use other variations.

This explains why different generators might understand the same prompt differently.

“A quiet night at a Tuscan restaurant with an old man sitting, drinking wine, oil painting”. On the left, Huggingface’s output. On the right, Craiyon’s output.

All of these images look like real paintings! Isn’t it awesome?

I know I gave you a lot of technical information. You might feel overwhelmed. But, honestly speaking, whether you understood all of it or not, it still remains fundamental to appreciate that there is no human magic trick behind AI-generated art.

There is no creativity. There is no sentience. There is no feeling. It is only math. Yet, that math brings to life something truly outstanding that can touch humans.

Sounds like a fun game. But where’s the revolution?

That was exactly what I thought when I first discovered AI image generators: they were a fun game. I only saw the possibility of finally picturing a flying cow in the desert.

Flying cow in the desert. 4 images
“Flying cow in the desert”. Generated using MidJourney.

That is the best flying cow I have ever seen and, probably, that I’ll ever see.

Admire this beauty. If it had been drawn by Magritte, Picasso, or any other notorious artist, everyone would have praised them for such inventiveness and imagination.

Now everyone must praise AI.

Jokes aside, the revolution is right in front of your eyes. There are endless possibilities: it’s all about imagination!

Imagine a child that wants to picture a flying cow in the desert. There you go! Children might benefit a lot from something like this.

Imagine a film director shooting the final scene of his apocalyptic movie that wants a sketch of the final shot: the “Earth exploding”. Try to draw it yourself. I know you don’t feel like doing it. Neither any director would. But with AI, in a few seconds, some ideas might come up:

Earth exploding. 4 images
“Earth exploding”. Generated using MidJourney.

Imagine a philosopher trying to describe what walking “into the unknown” is like:

Into the unknown. 4 images
“Into the unknown”. Generated using MidJourney.

Imagine a game designer, that wants to create the main character of a videogame. Something like a “mysterious man with a trench coat and fedora hat walking through a dark gritty street in the cyberpunk world with neon street signs illuminating the background”:

Mysterious man with a trench coat and fedora hat walking through a dark gritty street in the cyberpunk world with neon street signs illuminating the background
“Mysterious man with a trench coat and fedora hat walking through a dark gritty street in the cyberpunk world with neon street signs illuminating the background”. Generated using MidJourney.

I could go on forever. This is a spaceship for human creativity. I hope you see it too.

In the 19th century, with photography, you could capture anything you wanted and see the results after a few hours of exposure. It did not take long to change art and the world, forever.

In the 21st century, with AI, you can create anything you want — in seconds. I’ll say that again:

ANYTHING YOU WANT. IN SECONDS.

I guess it won’t take long to change things here as well. Don’t you smell the incoming revolution?

What doors has AI opened now?

AI image generators are a new thing. But some people have already taken true advantage of it.

The first person is me. Why? Because if I were to draw something for an art class as homework, I would copy everything from an image generator. That’s a guaranteed top mark!

As far as serious people go, “more than 3,000 artists from more than 118 countries have incorporated DALL·E into their creative workflows” states OpenAI. Check out their blog article for more info. You will be surprised how human beings have been able to reinvent these tools for their own benefit.

They are all spreading faster than Covid in 2020! And undoubtedly faster than photography in the 800s.

What doors will AI open in the future?

A revolution is not a one-time event. We are just at the beginning.

To give you an idea of where I think we are at the moment, recall that the first picture was captured in 1826. It was black and white. The first color photograph was taken by Maxwell (yes, the physicist) in 1861, 35 years later. That event changed things even more than the Niepce’s invention!

Now, the AI-ART couple is still at 1826 and is having lots of success. But we don’t know when we’ll start “seeing color” to achieve true success.

Who even knows a few years down the line what could potentially happen with animations or videos. They are just images stacked together, aren’t they? Imagine how social media might start implementing these tools in their apps for content creation!

I don’t know what you feel, but I smell a strong scent of revolution!

AI will open many doors, probably more than photography did.

What other doors will AI close?

Let’s be honest. AI is a newborn in our world. Art — taken in its most absolute sense — is not. There will be some inevitable clashes.

At the time of writing, AI will surely close my door which asks about prompts that include a face. Some of them were horrible.

No caption needed.

Everyone can fail. Even AI. But these models improve day by day, so faces won’t be terrifying for long.

Talking about serious problems… first of all, there are huge ethical issues. You can easily imagine that a bad person could try to create fake images of other people doing bad things, like with deep fakes. Those images will be pretty realistic because we have seen that these AI algorithms perform very well.

Something needs to be done about it.

We still have no true global regulations on AI — I say global because Europe actually has one. The effects of “not doing anything until something bad happens” could be catastrophic. History knows something about it.

By the way, this is also the reason why most of these technologies are not fully public yet: companies want to be sure that their products are used responsibly.

These issues go back to those of fake news as well. For instance, I could write an entire article on a new sea animal fossil discovered under the ocean, attaching just one single image to prove it:

A dead sea creature in a large fishtank

Is this image real? Is it AI-generated? I’m not going to tell you. This is what could really happen in the future.

Other issues could exist related to ownership. Who owns the picture? The writer of the text prompt? The downloader? The company that created the image generator? This is still unknown.

One problem, in particular, is dear to my heart: jobs will disappear.

This is the first time I truly understand the meaning of it. Why would a startup ever hire a designer to draw images for them, investing lots of money and time, when they could easily get an image of what they want in seconds for free?

I sincerely suggest you check out Marques Brownlee’s video to understand that the issue is not a joke. Maybe it is now. But in a few years, it will not.

Do I sound too pessimistic? I swear I’m not, for real.

Finally, is it good or bad?

I know it’s tempting to look at the cons! Most people do not cope much with such big changes. Especially if it’s about AI, for some unknown reason.

But someone once said: “We are sorry for the inconvenience, but this is a revolution”. I guess it was for a civil war or something, but the idea works here as well. That is, bad things could happen and there are aspects that will need to be readjusted. But when one door shuts, another opens.

Don’t be overly dystopian. The future of art is certainly unpredictable. That of AI too. But the odds seem to be going in a very clear and plausible direction, and we need to trust that it will be bright. You need to trust the process.

Try to be like Niepce. Try to see all of this as “the first uncertain step in a completely new direction”.

It is only by doing so that changes will become remarkable and go down in history. Just like it happened with photography, eventually, it will happen with AI too. That’s all it takes: trust.

I’m happy if you’ve come this far. I’ll award you with my favorite image so far: “eye of the universe”. Who cares if an AI did it. This is true art. And it is super cool. This is…

The Art of Artificial Intelligence

Eye of the universe
“Eye of the universe”. Generated using MidJourney.

Extras (optional)

  • To see more amazing pictures of the art of AI, check out DALL·E 2’s Instagram account.
  • Mark Zuckerberg has used AI image generators but in a unique way. Check it out on Instagram!
  • If possible, try to use the image generators yourself! It’s a lot of fun!
  • You still don’t know the text prompt of the cover image. Guess it!

--

--