AI Art…Ain’t

tl;dnr:  AI art is nothing but a search engine mash-up of collage and PhotoShop.

. . .

Now that we got that out of the way, let’s talk about what AI art is / is not, and how it will impact human creativity.

. . .

First off, humans have been using AI art or the equivalent for a long time.  An adult scrapbooking, a child slapping stickers of their favorite cartoon characters together, even Johann Sebastian Bach hizzowndamself play with the equivalent of AI.

Bach, you ask?  He invented a musical game where he wrote a handful of musical stanzas that could be played in any order, that order determined by a random toss of dice.

As far back as the Surrealist and Dadaist movements, artists would cut up old prints and recombine then in new ways, a technique still popular today.

The Dadaists also pioneered the cut-up technique of writing, in which stories where cut into separate paragraphs and passages, then those fragments recombined to create a new work.

In either the 1960s or early 70s a science fiction magazine (Analog?  I think so but can’t swear to it) published a story (by David Gerrold?) about a computer that started writing poetry (it turned out they set the margins too narrow on the printout out so they were only getting the center section of lengthier non-poetic texts).

MadLibs.

Already on YouTube are movies -- both short films and features -- made of hacked scenes from older movies (J-Men Forever by example) and video game engines.

A determined film maker could create an entire animated feature on their lonesome using a video game engine, commercially available AI voice programs for actors, and AI generated music cues.

So AI -- and what it does -- ain’t new.

What AI can’t do (at least as of this writing) is invent a new technique or style.

. . .

The first and most immediate worry by artists about AI is that it will cut into the number of illustrator jobs available.

Well…yeah…

That’s kinda the whole point.

It cuts into illustrator jobs the same what PhotoShop cut into illustrator jobs, the same way clip art cut into illustrator jobs, the same way printmaking cut into painters’ jobs.

Oh, jobs will still exist, but not as many and the entry bar will be set a bit higher.

Full disclosure: 
I’ll use AI generated images when I need to. 

Case in point: 
The artist robot above.

I swiped this from the NYTimes.

AI generated art is (currently) exempt from copyright since it isn’t directly created by a human being, so its fair game.

That’s the first speedbump in use of AI art.

You can’t own it.

. . .

The second speedbump is the current pure randomness of the images.

Even the most prolific AI prompt engineer needs to go through numerous close-but-no-cigar / dahfuq? iterations to find the handful of keepers.

Right now AI art programs filter their searches through their coders’ perceptions and values.

We’ll eventually get to a point where one can start calling up specific elements from specific catalogs, giving the prompter much more control over the final image.

Not much different from selecting / customizing a character at the start of a video game.

Add an overlay of art style filters and one can create a vast array of scenes and images with far more individual control than is currently available.

That will be a great toy.

But will it be art?

Spoiler:  No.

. . .

There are no unrecorded melodies.

Musicians Damien Riehl and Noah Rubin developed a program that recorded all possible 68.7 billion melodies in the twelve-note western musical scale.

They’ve released this into the wild; i.e., they’ve made all these melodies public domain.

Now, arranging / orchestrating / performing every possible combination of tempos / octaves / instruments /sounds has not been recorded or made public domain, nor has every possible combination of words as lyrics.

The specific expression of an idea -- musically / artistically / literary -- is still protected by copyright, but the basic tools, the building blocks are free for all to use.

You could create your own symphony just by stringing melodies together, adjusting tempo / key / orchestra to suit yourself.

So why don’t you?

. . .

“Now, Buzz, that’s easy for you to say. But what are you going to do when AI starts writing stories?”

Years ago I cooked up but never followed through on a scheme to create a bogus pulp fiction imprint and sell Kindle books through Amazon.

I would find public domain pulps westerns / mysteries / sci-fi / fantasy / horror on Project Gutenberg and strip out character / location / prop names (viz. Hopalong Cassidy would become Hero1, the Bar-20 Ranch Location1, six-shooter Weapon1, etc.).  Each work so mined would have its casts and locales described in that manner so that chapters could be taken from several different works of several different genres and mixed together.

I’d then search & replace Hero1, etc. with whatever names I would use for this “new” work, go over the text to edit out rough spots, and where needed add descriptions either of my own imagination or from travel or scientific texts found on Project Gutenberg.

By changing character and location names, I could take the same basic “new” text and publish different variants for different specific audiences, all the way from G-rated to NC-17.

I figured once I went through and did the basic editing to get my template chapters, I could crank out at least one “new” Kindle book a week.

You’re welcome to this idea if you want to try it; my interests no longer lay in that direction.  Eventually somebody will program an AI to do it and the bulk of popular fiction will be generated this way.

And as Ted Gioia points out, “The job of an entertainer is to find out what the audience wants and give them exactly that.”

The job of the artist is different.

. . .

“The writer's role is to menace the public's conscience. He must have a position, a point of view. He must see the arts as a vehicle of social criticism and he must focus on the issues of his time.” -- Rod Serling

Clearly Serling doesn’t mean writers are the only creators to bear that responsibility, and clearly by “of his time” he doesn’t mean only contemporary stories.

Anyone who ever saw The Twilight Zone knew there was no limit to when or where a story could take place.

Whatever the future of AI, it will not involve said AI offering original social criticism.

Whatever the AI expresses will be through the filter of the POVs of those programming it.

The majority of people who use AI art now and in the future will use it for personal amusement.

“Show me a picture of a duck in a tuxedo smoking a cigar.”

A sizeable number will use it for simple eye candy, much the same way I do with my fictoids and ephemeral art I use to illustrate my blog posts.

Attention grabbers, not works of deathless art.

A few will figure out how to incorporate AI into their own creative expressions, and how they will do that will depend on them individual.  It may range from something as simple as a visual idea to spark creativity (much the same as Magic Realism Bot on Twitter is a great source of story prompts) to some currently unimaginable application.

For the field I’m interested in -- creative fiction writing -- I can see a lot of possibilities.

Right now most consumer grade AI art consists of low grade still visual images, frequently crowded with disturbing glitches.

Fine if you’re doing something like Nancy’s Nightmares, not so fine if you’re striving for photorealism.

I can anticipate AI generated moving images in the very near future.

The great appeal of AI right now is that anybody can enter a few key words into the search bar and get a batch of images based on it.

As the systems become more refined, I can see scriptwriting for AI becoming a new field.

Writers create the basic story / characters / dialog but the consumer selects physical appearances and visual style.

There will be options to see the story exactly as the creator/s envisioned it, or just a pared down version one can play with.

The basic concept already exists in video games, it won’t take much more tweaking to get them to that level.

. . .

Is there a downside?

Of course. 
There always is.

A lot of folks making marginal livings now as illustrators are going to be either pushed out of the field entirely or reduced to semi-pro status.

For some, this will be a good thing: 
They can focus on what they want to create, not what somebody pays them to create.

For others it will be heartbreaking.

I genuinely feel for those souls.

Success in a creative field requires large measures of skill / talent / luck.

AI is going to level the skill portion of the playing field.

Actors / singers / dancers / performers will also face career challenges.

We recently saw the sad spectacle of Bruce Willis grabbing as many last movies roles as possible before his aphasia rendered him unable to work.

I don’t fault Willis’ for that; bravo, in fact, for persevering.

But we’re already able to recreate real people in CGI so well that the CGI image can pass (at least briefly) as the real person.

Right now we’re using it to put actors’ faces on CGI characters in dangerous stunts or insert them in brief scenes where their schedule (or death) prevents them from appearing in person.

We can and will soon cast movies entirely with CGI casts of real people, living or dead.

That’s an option audiences around the world will want, replace an actor you don’t like in your favorite film with one you do.

Savvy actors and models are already licensing their image and voice for future work long after they retire or die.

If a producer wants to make a movie with a now deceased superstar, it’s gonna co$t.

But what if a consumer wants to do it in the privacy of their own home?

How ya gonna monetize that?

The style can also be adjusted.  Parents can let their kids watch a G-rated James Bond movie after dinner -- No sex or violence, just comical villains, chases, and narrow escapes -- then later in the privacy of their bedroom watch the 007 epic’s NC-17 version.

If they so wish, they can adjust Bond’s sexual orientation to something that more suits their fancy.

Don’t look at me like that, there’s already a huge amount of CGI porn out there.

. . .

On the third hand (to quote Larry Niven & Jerry Pournelle), CGI characters in AI entertainments will open up interesting opportunities for performers.

Certain professions are more impressive live than in CGI. 

Anything involving real physical motions -- dancing / stage magic / juggling / mime -- will be more impressive than its AI generated counterpart.

Even the most realistically rendered CGI magic trick is just animation, but somebody making an object disappear in front of your eyes is real.

Stand up comedy / poetry slams / staged performances will find audiences that want to see and hear real people doing and saying real things. 

There can be a resurgence of regional and community theater, everything from amateur theatrics to store front little theaters to big outdoor pageants.

We’re seeing right now a new wave of hobbyists and home crafters making things for the sheer joy of it, selling them only as an afterthought either at local bazaars or online through e-stores like Etsy.

Those are real and unique items, not mass produced product.  As many people who enjoy wearing brand names, I suspect there are as many wanting something unique to them.

. . .

AI is going to alter our view of a lot of things.

That’s the nature of the universe: 
Nothing remains the same forever.

It’s pointless to be upset at the rise of AI in all areas of life, and it’s fruitless to tilt at the AI windmill and pretend it’s a fad that will pass.

Change isn’t coming.

Change is already here.

© Buzz Dixon