In the wake of Bandcamp’s decision to ban all AI-generated music from the platform, it seems timely to argue yet again and possibly ad nauseum for human input into art. To be up front, I do not want to live in world where AI is the only thing making ‘art’.
The Burning
I’m a big fan of music. Not a musical note in my body or soul, but music has long been the soundtrack of my life through the new jobs, bitter divorce and the endless idiot mistakes I have made, through which I have hurt a great many people around me. People have come and gone, for the most part. Music has stayed. Music has guided. Music has healed.
On a social network, I was recommended a Christian SKA punk group because a friend liked them and thought I, as one sharing their faith, would also enjoy their music. Imagine my surprise when I found the band was not only non-Christian, but also it was not even human. On the platform where the band was selling their music, there had been no label or hashtag to indicate the music was made by generative AI.
I did what any normal person would do. I hid all of the albums I had bought from my collection on the platform (I don’t want followers buying this rubbish because I did and, more honestly, I am ashamed I bought four albums!). I also reached out to the man behind the music and, to his credit, he wrote an extremely lengthy reply. From a theological viewpoint, he is very much into the occult, shamanism and other things which most Christians would, at best, find questionable. The music was tagged as “Christian SKA”, a point of contention. He also went to some great effort to explain his justifications for using AI, his status as a real musician, and how the resulting music from the AI really did embody him and his style. He also said he had spent months on creating and refining the music. Since sending that email, the same guy has punched out another two or three albums in a couple of months. Needless to say, I got well and truly burned here.
And this dragged my hypocrisy, kicking and screaming, into the glaring sunlight of my self-professed ethics and morality. Music or images? What is the difference? Using AI to search for something, is one thing. Using it to create “art”, whether music, images, or words, is something completely different.
The Thinking
One “musician” uses generative AI, and they go by the name Jack Righteous. In his explanation on his blog, he likens the use of AI for music to a musician using a Digital Audio Workstation (DAW) to produce the final work, among other things. Jack Righteous attempts to place AI in the same category as other tools like the synthesisers and DAW’s a musician might use to create their sound and get it out there. I have some serious problems with this and you can probably see where I am going with this.
Firstly, an imaginative scenario can show why using AI does not make someone an artist or a musician. Let us say I have used the image above as a header image for my blog. I clearly did not draw this or use any digital tools like PhotoShop to paint this. Rather, let us imagine I commissioned Erik Shoemaker, (the actual creator of the picture), to make this for me. No one would consider me to be the “artist” in relation to the artwork. Shoemaker is the artist. I contend the same goes for anyone using generative AI to create a piece of ‘art’. That person is not an artist. The AI has taken their “commission” in the form of prompts, just as I would have supplied to Shoemaker, and it has generated an image based on those instructions. The same scenario holds true for music. Those who use AI to generate music are not musicians; certainly not any more than I am. They have supplied a group of instructions (prompts) to a programme and that entity has generated an outcome. The only difference in these two scenarios between the musician, artist and the AI is the humans produce superior work which can actually be referred to as art.
Secondly, to classify generate AI as another tool, along side synthesisers and DAW’s is borderline dishonest, or wilfully misleading. When a musician uses synthesisers, they have either written the music themselves or they are using someone else’s music. Either way, a musician is creating the music on the go through a machine which makes different notes in response to the keys, strings or other manipulations the musician performs. Paul Henry Beaver Jr, in 1967, was one of the first to use the Moog synthesiser in the Doors’ song, ‘Strange Days’. Beaver pressed keys. Beaver created new sounds. To claim an AI user is somehow in the same vaunted group as Beaver is insulting.
In the case of DAW’s, the categorisation is hardly less tenuous. Using a DAW requires a musician to have a musical piece already written and composed. It might be theirs or it may be someone else’s piece. Either way, someone with a specific skillset has spent time writing, refining and revising the piece of music before they even start mucking about with a DAW. The contrast with generative AI makes Uluru look like a pebble. Generative AI, quite obviously, has a vastly different beginning point. People are not bringing their own, pre-written music to the programme. It is the AI itself which is creating the music. The human, apart from supplying some instructions, plays utterly no part in the process of writing the music at the beginning. Even should a musician use AI to refine their music, that is something they have written, the criticism still stands.
Generative AI, DAW’s and synthesisers do not belong in the same category.
The Buzz Word
One of the most exciting aspects of AI in music is its potential to democratize the creation process.
In the above quote, Jack Righteous attempts to position generative AI as something which makes music creation more accessible and fair. This is perhaps his strongest argument. It is not a good argument, but it is about best he has. “Democratizing” something taps into Jack’s audience’s culture and political landscape. Being North American, Jack is from a nation which values democracy highly and takes no small pride in the system they have. At first, the use of this word sounds good and positive. At first.
AI does not democratise anything. First, the word itself, according to the Cambridge English Dictionary means (and I quote):
to make countries or organizations use democratic ways of making decisions.
The word “democratise” has absolutely nothing to do with art creation. It is a word which refers to decision-making. By devoting an entire section to “democratizing music creation”, the only objective Jack Righteous has achieved has been to throw in an irrelevant, buzzy word in the attempt to give more substance to an argument.
Jack Righteous seems to be aiming, rather, at accessibility to the creation process, arguing AI makes creative expression more accessible. That is, specifically the creation of music. He fails to make a justifiable argument in this point, as well. What he is really arguing is generative AI allows people to by-pass years of training and practise, using a quick and easy shortcut to so-called “musicianship.”
There are a whole host of skillsets for which the last thing we need are quick and easy by-passes to real effort. Imagine you need open-heart surgery and you go to the hospital for your operation. The nurses prepare you, the surgical team coach on what will happen and you are feeling pretty good about the whole process. You are feeling quite safe, given the circumstances. The outlook is positive. Things are looking great, except for the hospital gown, but you have other priorities on your mind. When you are wheeled into the theatre, you are puzzled. You see none of the surgical team there. Instead, you are greeted by my face, replete with surgical scrubs and you notice I have a computer on my tool trolley. It is helpfully opened to ChatGPT and some instructions on how to perform an open heart operation. This ridiculous scenario highlights some skills should not be “democratised”. Surgeons, pilots, nurses, psychologists, pharmacists, lawyers, scientists and more cannot just take the generative AI shortcut to professional success. There are exceptionally good reasons why these professions spend years and years in training and education, studying their fields. The last person you want doing any of those jobs is Bryan Beal armed with ChatGPT. If this is true of the skills listed above, it follows that art is not necessarily something we want treated in the same way. I would argue that it should never be “democratised” in the way Righteous argues.
As alluded to above, I am no musician, and there is a very good reason for that. I am lazy. I have not invested the years and decades of practice, learning and grind to acquire the skills necessary to call myself a musician. Art is not something that should be “democratised” so people can take gargantuan and unethical fast tracks. By necessity, such an approach cheapens and denigrates the art of those who have put in the effort and made the sacrifices needed.
This is Political, Baby!
There is very strong political shift happening in art and it is something we should be extremely worried about. It has been happening for some time, but generative AI exacerbates the shift and the issues surrounding it.
The real artist or musician has direct and total control over the means of production. The musician, by controlling directly the writing process, controls the means of production in that they control themselves and have no limitations on what they can say or play. ELO can write and play whatever they like. In creating his visual art, Erik Shoemaker can do and make whatever he likes. He is the artist and the only limitation he has is his own imagination. In both of these cases, the means of production itself rests with the musicians and artist. There is no other component.
The same cannot be argued for generative AI. The user (they are not an artist or musician) surrenders the means of production to the company or corporation which develops and sells access to the algorithm. As an example, you could attempt to insert a prompt including the words “big boobs” or “blood gushing from a gaping and open wound” and Nightcafe (an AI image maker) will tell you those terms are forbidden. Erik Shoemaker would have no problem creating an image of a Amazonian warrior with “big boobs” using a sword to slash open a gaping, blood gushing fountain of a wound in her foe. That is because Shoemaker controls the means of production. The AI user does not.
Rather than “democratising” artistic creation, as Jack Righteous uses the word, by using generative AI, users are doing the exact opposite. They are surrendering control of the artistic process to corporations like OpenAI, Google, Microsoft, MidJourney, SunoAI and more. This sets a dangerous precedent of corporate control over the artistic process. My example of the buxom warrior princess is an extreme example, but it prompts us to consider wider socio-political issues to which art may have the need to speak. Corporations can implement any restrictions they wish to. The generative AI platforms are privately owned by the shareholders or investors providing the capital.
Some of Elon Musk’s capital used in his acquisition of Twitter came from the founders of Oracle Corp., one of whom is Saudi Prince Alwaleed bin Talal. How would Prince Talal respond to a news service using Grok (the X/Twitter AI) to produce material critical of the Saudi Royal family? Erik Shoemaker and Sting can say whatever they like through their art. The user of AI is limited by the socio-political concerns and views of the owners of said AI.
At times, artists need to offend people to make a point society needs to hear. At other times, artists made need to push the envelope in order to get a message out there. Andres Serrano did just that in 1987 when he created Immersion (P*** Christ) and caused a storm of controversy. Although Serrano reportedly remains a believer, his artwork takes aim at organised religion on a number of different levels and about a number of different issues. It is important our society preserves such freedoms as necessary for artistic expression. As a devout Christian, I personally found Immersion a little uncomfortable, but I utterly support the right of Serrano to offend my sensibilities.
Such freedoms can be greatly restricted by corporations and often will be, but based on very different reasons. A company’s most important priority is profit. That is the beginning and the end of their existence. Anything which threatens profitability is more likely to be censored by the corporations with control over the means of production. In the extreme case of Gab Social, GoDaddy dropped them, and PayPal and Stripe followed suit. To think this was purely about the ethics behind Gab Social’s vile content is naïve. This was about profit. Facebook (now Meta) has a long history of allowing disagreeable and offensive content to remain. If you are engaging with a Holocaust-denying troll, you are spending time on the network and that means profit. Censorship is based on what makes money. And that is a dangerous thing on which to base limitations on artistic expression.
The Swansong
AI users are surrendering their artistic expression for the sake of a rapid shortcut to so-called “artistic skill”. To claim generative AI is not a threat to artists is to deliberately ignore what has been happening in the entertainment industry. Most people should remember the 148-day strike by writers in Hollywood. A fast-track to avoid real effort is an extremely questionable reason for investing trust in a corporation, an entity with a long history of shafting its own employees, the community and anyone else except the parasitic uber-rich.