Ultimate Guides

AI Art: The Ultimate Guide for Fine Artists

Why fine artists should explore, understand, and perhaps even try making art with artificial intelligence

Artificial Intelligence is increasingly getting attention, in areas as diverse as writing, games like Chess, and social media algorithms. Recently, we’ve been seeing more art being created with AI–yet another example of technology entering into the art world.

Whatever your gut reaction to this as a fine artist, we recommend at least understanding the basics of what AI art is, before making up your mind. Who knows, you might even decide to test it out!

So here it is: The Ultimate Guide to AI Art!

This AI-generated painting sold for $432,500!
Source: https://news.artnet.com/art-world/artificial-intelligence-art-history-2045520

Should You Worry about AI Art in Your Art Career?

AI Art is here. 

It seems to be the “next hot thing” in the art world. People are talking about it. Christie’s is auctioning it. Students within the Professional Artist Institute are testing it out.

Should you be worried about it?

Is AI here to replace traditional fine art? Will fine artists be out of a job due to AI, like self-driving cars threatens to do to truck drivers? Should you be learning to make AI art?

Don’t worry!

There’s an old saying, generally attributed to Plato, “There are two things a person should never be angry at, what they can help, and what they cannot.”

Like social media, and NFTs, AI art is here to stay, grow, and find its place. So there’s no need to worry. But just like AI, as artists, we too need to evolve.

If we look at the steady march of technologies, from cars over horse-and-buggy drivers to typewriters over scribes–it’s clear that people are never replaced, as long as we learn to adapt.

From a positive perspective, perhaps AI art is yet another tool or technique for fine artists to learn, explore, and create with!

Sound familiar? It’s like a website became the new artist portfolio, Instagram became the new business card, digital artwork and NFTs are finding their place, and now AI art is here to disrupt the art scene again. Certainly this is nothing new–at one time oil paint was a new technology, as were pastels, but it certainly feels like the pace is accelerating!

3 Reasons Not to Worry about AI Art

Technology Takes Time

Even if AI Art were to grow larger than hand-crafted artwork (which seems unlikely), this would likely take decades to occur, slowly. The best you can do is learn about it, keep an eye on it, and simply carry on creating art as you do. 

Don’t worry 😊

There will Always be Collectors for Traditional Fine Art

Digital art has been here for over 40 years and hasn’t overtaken “analogue art” yet! Sure, it’s finding its place–like with NFTs, but good old-fashioned art has well and truly kept its place too!

Don’t worry 😊

You are a Unique Artist

You are unique as an artist. Therefore you are irreplaceable!
Let’s be honest, there are hundreds of millions of other artists out there.
Yet, you still have your individual style, niche, and importantly–personality!
No AI is going to do your work for you (actually, maybe that could lighten the load…)

You are on-one-a-kind. Irreplaceable.

Don’t worry 😊

Source: https://collections.vam.ac.uk/item/O499573/untitled-drawing-cohen-harold/

What is AI Art?

Let’s rewind. What is AI Art?

For at least 40,000 years, the only known source of artwork has been humans (and perhaps Neanderthals). That was, until the seventies…

In the late sixties/early seventies, American computer scientist Harold Cohen created the first-ever AI drawings and paintings. You can see one of his first images above. While arguably not a masterpiece, it sparked something that continues to evolve today.

It’s kind of like training a puppy to play fetch…

Just like the AI known as Deep Blue was trained to play Chess, until it beat the Grandmaster Gary Kasparov in 1997, computers can also be trained to create art.

Although to be fair, few would argue that art is more complex than a purely mathematical game like Chess. So it’s taken a few decades to merit the attention of the art world. 

And here we are today, with the first AI-generated portrait reaching $432,500 under auction at Christies. The name of the portrait?
Portrait of Edmond de Belamy, from La Famille de Belamy (2018). (Image below).

If you’d like to recreate this, here’s a hint: “𝒎𝒊𝒏 𝑮 𝒎𝒂𝒙 𝑫 𝔼𝒙 [𝒍𝒐𝒈 𝑫 (𝒙))] + 𝔼𝒛 [𝒍𝒐𝒈(𝟏 − 𝑫(𝑮(𝒛)))]”.

Portrait of Edmond de Belamy, from La Famille de Belamy (2018). Courtesy of Christie’s Images Ltd. 

Definition of AI Art

Let’s pause a minute to define what AI art is exactly. 

The simple version is, “AI art is computer-generated art”. 

The more complete answer is, “AI art is artwork generated by a computer that learns with rules, heuristics, trial-and-error learning, and being fed with good examples.

And as the old computer science adage goes, “garbage in, garbage out” (GIGO).
This simply means that to learn well, the computer needs to be “fed” with good examples of art. Like in the example of Portrait of Edmond de Belamy, where the creators fed the system with a data set of 15,000 portraits painted between the 14th century to the 20th”. 

Sounds a lot like art school! Study the masterworks, and learn to paint like the masters!

Wait, What’s Artificial Intelligence?

Okay, we’ve all heard the latest buzzword, “AI Art”, but… what does that even mean?

First, let’s start with AI. 

Artificial Intelligence comes in two flavors: general and specific.
(Okay, if you’re an expert in AI, you’ll distinguish many types of AI, but the simple version is, there are two kinds).

General Artificial Intelligence

General AI is what you see in Sci-Fi movies like The Matrix, 2001: A Space Odyssey, or The Terminator. Basically, it’s the idea that computers/robots will become intelligent like humans and think for themselves–with good or bad consequences!

Tl;dr General AI does not exist… yet (as far as we know).

Specific Artificial Intelligence

Specific Artificial Intelligence is different. And importantly, it’s already here!

There are many kinds of specific AI.
For example, there’s the AI behind each social media platform’s algorithm.
There’s the AI suggesting what to order on UberEats.
And there’s the AI that writes articles just by inputting a topic and a few prompts (sorry–this article was not written by AI!).

The History of AI Art

Where did AI Art come from, and why are we only hearing about it now?

There’s a general trend with technology–the first iteration isn’t pretty. Only a few “early adopters” will use it. The first Smartphone was released in 1993, and was clunky like a brick. It took another 14 years, until the iPhone was released, and the full power and elegance could be seen.

Similarly, the first AI was, well, ugly. Okay, it’s all subjective but, we can all agree that it lacked complexity. Certainly, it was incapable of realism. Importantly, it could not create in color.

1843: First Concept of Computer Generated Art

Perhaps the first idea of artwork, generated by computers was by Ada Lovelace. Way back in 1843 in the earliest conception of computers, which she termed an analytical engine, Ms. Lovelace predicted how such a machine might be used to compose complex music, and produce graphics. How prescient she was!

1950: Earliest Description of Artificial Intelligence

Alan Turing is known as the father of computer science. In 1950, he developed the Turing Test, also known as Imitation Game, the benchmark test for whether a machine’s behavior can be distinguished from a human.
Perhaps this will become the hallmark test for the quality of AI art… 

1973: First AI Art

Classically-trained artist Harold Cohen also studied computer science. Fusing these two disciplines, he begins training an AI he names, AARON. While its initial creations are child-like blue line drawings, he continued to train and refine AARON for the next 40 years.
In 1995, it produced its first color image (below), showing a marked increase in “intelligence” from its early beginnings.

Source: https://computerhistory.org/blog/harold-cohen-and-aaron-a-40-year-collaboration/

2018: AI Portrait Auctioned at Christie’s for $432,000

Shocking, right?

Almost as shocking as the NFT that went under the hammer for $69,000,000 in 2020.

It’s worth noting that there seems to be a trend here–the “first of a kind” gets valued very highly. Later, it seems that a law of diminishing returns applies…

Importantly–unlike the early beginnings of AI artworks, this portrait painting is actually artistic, albeit minimalistic. It would be difficult to know that it wasn’t painted by a person.

Somehow, the quality of art, the auction at Christie’s, and the high valuation, appear to have established the place of AI in the art world.

Today: Simple Tools to Create AI Art

All new technologies start highly complex, and become more user-friendly over time. Art made with artificial intelligence is the same. It began as something for computer science wizzes, and now anyone can learn it within hours. Of course, to be fair, it may take some practice to gain mastery of–but you can get started quickly.

Below we introduce a “quickstart guide to making art with AI”.

AI Art for Fine Artists

Artists are natural explorers. We learn, evolve, and refine our skills for a lifetime. In part, this is responding to our environment, and partly it is learning new skills.

If you’d like to try your hand at creating AI art as a fine artist, here’s a brief way to get started today.

Okay, so you’re curious–how to go about trying out this new thing?

The AI Art Stack

Please understand that there are two layers to AI art.

Underneath, there is the artificial intelligence program. Currently, the most powerful and popular of these are known as DALL·E 2, Stable Diffusion, and Midjourney. To use (or, “talk to”) these directly, you need to “prompt” the AI with either text, or images.

A prompt could be as simple as, “Jim Carrey in circus”. This will generate several (likely funny) images. From these, you can choose to generate further variations based on the one you like best. Like, for example, if you prefer the image in which Jim Carrey is ringmaster, you could opt for further variations on this.

Images generated with Stable Diffusion

On top of each of these programs, various platforms can, and are being built. These tend to be more user-friendly, and more specialized (like StockAI.com, for stock images generated with AI, or InteriorAI.com, for AI generated interior design–both of these are being built upon DALL·E 2).

Top 3 Programs for AI Art

While this space is evolving quickly, the top platforms for AI art are currently:

DALL·E 2

The next iteration of DALL·E, this is being developed by Open AI. Its strength is in working with natural human images, like the very challenging subject of the human face.

To try it out, you’ll need to sign up and wait–due to heavy demand, access is being rolled out over time. We suggest requesting access now–you may want to use it down the line.

Midjourney

Each AI has its own flavor. Midjourney’s often appears futuristic. It also excels at abstract. To glimpse just how powerfully it can interpret the name of an emotion, take a look at these examples.

Access is in the open-beta stage. Best to apply here now, and try it out when possible. 

Stable Diffusion

Unlike the previous two AIs, Stable Diffusion is in public release. Since August 2022, you can fully use it today. For simply this reason, this is where we recommend starting. 

As said above, many tools and platforms can, and will be built upon this. To name two household names, both Canva and Photoshop are building tools with Stable Diffusion.

It’s 2022–the future is now. Below is the quickstart guide.

Quickstart Guide to Making AI Art as a Fine Artist with Stable Diffusion

Okay, you’ve read this far–let’s do this! 

This is easiest done on a laptop/desktop. Maybe possible on mobile, but challenging.

3 Steps to Generating AI Art with Stable Diffusion

  1. Take a look and search what is possible to create on Lexica.art. Don’t hesitate to make your searches highly specific.
  2. Inspired? Great! Now register an account at Dream Studio
  3. And… play!

Wait–it’s that easy? 

Yes! But… to really refine this you’ll want to master two aspects:

  1. Learn to tweak the image parameters like dimensions, “cfg scale”, and steps.
    These parameters are in the panel on the right-hand side.
  2. Read about, understand, and test “prompts”.
    This is the difference between a first time user, and an AI artist.

Your prompts can start with a “raw prompt”, like a subject. “Pineapple”, for example.
You can then add a style, like “hyperrealism”.
And even adding the name of an artist, in whose style, you’d like to imitate. For example, “Paul Cézanne”.

Here’s the results of a progression from “pineapple”, to “pineapple. Hyperrealism”, to “pineapple. Hyperrealism. Paul Cézanne.”

There is further subtlety to prompts, but this is the quickstart guide to get you going.

Who Owns the Copyright to ‘My’ AI Art?

You may be wondering who owns the copyright–you the artist, the AI company, or…?
After all, who created this?

In general, it’s still being debated. However, the U.S. Copyright Office did recently reject a copyright request for an image generated through artificial intelligence, citing that it lacked the “human authorship” necessary for protection.

Overall, it’s not clear at this point, but it appears that AI artworks do not grant copyright protections.

The Future of Art with Artificial Intelligence

When something becomes too simple or inexpensive, it becomes a commodity. Just think of photography. The earliest photographs–dating back almost 200 years, were highly prized and valued. It’s a different situation today, with powerful cameras ubiquitously tucked into billions of pockets worldwide. Pictures are a dime a dozen on Instagram.

And yet… great photographs and photographers continue to be celebrated. Some even prefer to work in analogue over digital.

So?

So when something becomes commoditized, quality shines forth even brighter!

This has two clear implications for fine artists.

Firstly, in a world where anyone with a computer will have the tools to generate AI art, there will be a lot of ‘noise’ from millions of amateurs. The few who perfect their skills and bring an artist’s eye to the process, will stand out from the crowd.

Secondly, if we do see a deluge of computer-generated art, traditional, hand-crafted, artwork will become ever-more appreciated and valued. Just think of the distinct perceptions of original oil paintings over prints.

For these two reasons, ‘analogue’ art will always have its place, be valued, and is not going anywhere!

An example of an image generated by DALL·E 2, with the prompt, “a painting of a fox sitting in a field at sunrise in the style of Claude Monet”. Source: https://openai.com/dall-e-2/

AI Art and the Professional Artist Institute

At the Professional Artist Institute, we continue to learn about emerging trends and tools in the art world. We’re committed to keeping our finger on the pulse, to keep you up to date. Recently, we’ve been hearing more and more about AI art. Some of our students have even reported that they are trying their hand at it. 

Although AI Art may not be every artist’s cup of tea, it makes sense to understand its place and importance in the landscape of the art world.

As we watch this space closely, we’ll evaluate whether this technology merits a module within our flagship program–the Professional Artist Accelerator.

Read More