Does an app have experiences and memories you can relate to? Does it have family and friends? Does it have any life outside your phone? Does it understand what it’s doing or why? Art requires a human. If there’s no human, it’s not art.
Is software even alive? Does it breathe and grow like a plant or animal?
You’re probably thinking, no. Maybe even that these questions are kind of ridiculous.
They matter because humans are the only thing we know of in the universe that makes art. Elephants may paint, and birds sing, but those natural abilities are not experienced as art by most humans.
Sure, a computer has a memory, but only about one millionth of what a human has, and a much more restricted kind of memory. What it remembers is letters and numbers, not multi-sensory experiences. Virtual reality can recreate experiences, but they require a human to perceive and make sense of them.
Further, very few people think that their computers are alive. There is no evidence of life that doesn’t use DNA. Software can communicate, but there’s nothing alive in the silicon and wires, despite the impressions of consciousness attributed to them by human observers. Someday we may have sentient robots like in science fiction, but we’re not close to that today.
So this is the big question. If computers aren’t human and aren’t even alive, then how can they make art?
They can’t. What they can do is collect art from all over the internet and mash it up in various ways, some more intensive than others. Some is public domain and much of it is under copyright.
Using or changing art without permission of the copyright owner is illegal unless the resulting work is significantly transformative. That’s why there are legal challenges to AI art. But it’s complex because we’re not clear if it’s the programmer, user, or both that are responsible in current law. Currently you can’t register an image made by AI with the copyright office, and AI makers are being sued.
For the moment, let’s assume there’s an accommodation for copyrights.
Let’s look at what happens when entirely machine made images or sound are labeled art.
We can get clues by looking at what happened other times when machines were used to replace human skills in music.
For thousands of years, if you wanted to listen to music you asked someone who could read and play music or played it yourself. The invention of recorded music allowed anyone to listen to music without having to learn how to play. An important exception is in churches, where everyone reads music from the hymnal.
So, recording caused many people to lose music literacy, but mass produced recordings helped them be better listeners. They had access to the world’s finest performers. And live performers could record. So there was a balance of benefits to those harmed by the new tech.
Then, in the 60s and 70s, samplers and drum machines began to replace live musicians. This tech put musicians out of work, without new jobs for them. The new trend changed public taste. Mass audiences were trained to accept the crude sound of fake strings, or the mechanical beat of a fake drummer. This made people worse listeners.
The sound and composition were less human, but still marketable.
The automatic trend continued with Autotune in the 90s. This allows anyone to sound like they can sing, similar to programming a drum machine or a sampler to make it sound like someone knows how to play drums or violin. The problem is, the effect is poor. The detail and expression of a voice are largely removed, and the flow is stiff. It’s more a trick or effect. But it’s everywhere.
Again, audiences are trained to accept this lower quality product. Instead of focusing on the quality of the composition, or of the performance, there’s more attention on production values, such as how clean and clear the recording is.
Now, AI music can complete eliminate the requirements for a human to do anything. Everything from composing to production are handled by software. Prompts can be autogenerated and the AI just cranks.
And that brings us to the most important concepts of all: Talent and Creativity.
People have been observing, studying, and writing about these abilities for thousands of years. There is no clear answers yet, they’re mysterious phenomena. Partly it’s genetic, and partly early training, but for our purposes, we’ll say that musical talent is the ability to quickly and accurately learn the building blocks of Music, including detailed pitch, rhythm, and timbre, and then applying them for expressive purposes. This is the definition of being an artist.
Now and for the foreseeable future, no computer or software has talent. The subtle, expressive nuances of a human performer simply can’t be learned by a neural network. By the time someone figures out how to train it we’ll already have sentient AI (with a new set of problems and opportunities).
It’s what’s known as an emergent property of living systems, the prime example of which being humans.
If a human user wanted to take AI music and add in the subtle and not so subtle elements that are missing, it would take time, and someone who themselves had talent and applied it to the task. Most users won’t care enough to do that or lack the talent, instead, flooding the market with dehumanized sound masquerading as music.
Then there’s creativity. Another subject of long study that remains mysterious. But for our purposes we’ll say creativity is imagination and originality applied to a specific task.
This is another emergent property of living systems.
Combining bits from pre-existing works of art can be a part of creativity, but it’s clearly not the whole thing. First off, originality is minimal. And I don’t think anyone is claiming their software has an imagination. (But maybe someone!)
Again, a human with imagination could try and add that into an AI product. But that would take further time and skill. That most people don’t have or would want to do.
Maybe it could be a tool for people who already have talent and skill. But for the most part, that isn’t what’s happening.
There are over 100000 songs uploaded to Spotify alone every day. That’s going to increase exponentially now that AI is able to generate cheap songs quickly, producing an endless stream of crappy, mediocre work that any ignorant or money addict publisher can promote to billions.
How can the market decide if the choices are so limited? Many companies will need to aggressively promote new human artists, or producers with machines will replace artists.
The whole idea of a natural talent becoming a master artist is to produce art the world needs — someone whose mind thinks in music, or painting, or cinema, or architecture, or dance, someone who lives in the world, and responds. That’s why talent evolved: to bring necessary art to the world. When fake art is mass produced it dilutes the and tends to crowd out necessary content that humans create for one another.
And most of those AI songs won’t have the necessary talent and creativity in them to make them worth listening to. But public taste has been systematically lowered by previous generations of automated music production. People can be trained to like things that make them sick.
This means it will be harder than ever for gifted artists to be heard amongst the gigantic throng, so whoever can afford the best marketing campaign will have an even bigger advantage than they do today. This will have the knock-on effect of discouraging budding artists of every stripe from even beginning their journey. If you’re just starting out, seeing and hearing the high technical quality but soulless torrent of art and music being produced by AI can be soul crushing.
We just don’t know enough about what makes a good artist let alone a great one to be able to quantify it and program it properly, but we do know how to lower peoples expectations and how to market mediocre material.
There are those who say AI is just a tool. Well, some tools are a lot more dangerous and require a lot more skill to use. You need a license. But AI has no such requirements. Not yet. People in congress need to get educated. Imagining that writing a prompt is enough to call an AI sound file “art” is crazy, it would show extreme ignorance of what it takes for artists to make good art.
We need to regulate software. Especially when it’s potentially dangerous. The question of how we do this arises, and lawmakers typically have trouble with the understanding of even simple technology, and as a result tend to write bad law.
Nature evolved talent and creativity over millions of years. And now some people are trying to replace it with their software. No thanks.
I believe there is no substitute for reality. True art requires a talented person who sees what others don’t, who can respond, and innovate. We should be very careful about exposing people to large amounts of oversimplified, computer-made artifacts. Humans make art and it takes talent and work. There’s no shortcut. You need a life to create Art.
I say we never copyright AI art1, and do what we can to prevent thoughtless or unscrupulous people from attempting unfair competition — or damaging our soul.
AI art can be fun. But code is ammoral.
An overview of studies on the nature of music talent, The Genetic Basis of Music Ability
The same people who do ChatGPT have a good visual description of how it works.
1 The first copyright for an artistic creation produced with an AI art tool (Midjourney in this case) was awarded to Kris Kashtanova for her graphic novel Zarya of the Dawn. She successfully argued that the work was not simply AI generated, but was created using AI as a tool, with direct human guidance from her.
Despite popular misconception, the US Copyright Office has not ruled against copyright on AI artworks. Instead, it ruled out copyright registered to an AI as the author instead of a human.
David Raiklen wrote, directed and scored his first film at age 9. He began studying keyboard and composing at age 5. He attended, then taught at UCLA, USC and CalArts. Among his teachers are John Williams and Mel Powel.
He has worked for Fox, Disney and Sprint. David has received numerous awards for his work, including the 2004 American Music Center Award. Dr. Raiklen has composed music and sound design for theater (Death and the Maiden), dance (Russian Ballet), television (Sing Me a Story), cell phone (Spacey Movie), museums (Museum of Tolerance), concert (Violin Sonata ), and film (Appalachian Trail).
His compositions have been performed at the Hollywood Bowl and the first Disney Hall. David Raiken is also host of a successful radio program, Classical Fan Club.