Machines & Dreams
What we learn and what we lose when we work with machines
Two-part reflection on the 1. magical, symbiotic relationship between creators and machines and 2. the ever-shrinking rights of creators in the digital age
The name of this Substack is a phrase I have been kicking around since I was a teenager. At that time, I imagined the push and pull between our human desires and the cold, sometimes mechanistic world as fraught and in constant tension. Now I see our relationship with machines and all tools as much more symbiotic, even as the rewards for creative collaboration with machines have diminished.
Just as the iPhone has replaced the watch, camera, typerwriter, etc., any piece of recent technology geared toward creators has replaced many other tools as well. As an artist I am elated by the fact that you can essentially fit a five million dollar recording studio flush with consoles, tape machines, amplifiers and all sorts of sound-making toys into a piece of apple software that now costs about two hundred dollars. And in the visual arts, the speed and agility of Adobe Photoshop is similarly unmatched. These software perform creative feats that would have taken incredible amounts of time and money (and staff) just a couple decades ago. Remarkably, Generative AI is quickly making creative production even easier and faster. Generative AI will also likely accelerate the declining value in creative production.
There does seem to be a relationship here between the increasing ease and low-cost of creating and the diminishing value of those works of art in the market. On one hand, I am tempted to conclude that yes, of course a song that cost a few hundred dollars and a few hours to make is going to make less money on streaming services than a song that cost thousands of dollars and a week of studio time would have made on the radio. But this view is only convenient for the silicon valley owners of these technologies; contemporary technologies promote the idea that value is in the software, in the dissemination of it, in the ease and speed of it all—rather than the real human creative vision of individual creators, artists and technicians. AI is poised to take this view to the extreme, and I worry that we are unlikely to ever come back. But before thinking more about what we are losing, let us celebrate what we have gained.
The magical, symbiotic relationship between creators and machines
At nineteen I was listening to an early version of a song I had written a year or so prior. The track had been recorded and heavily edited by an established record producer using an early version of Pro-Tools recording software. Recording with computers was still new at the time, and record producers were quickly realizing how easily it was to fix pitch and timing errors. Listening back to the recording of our song, my drummer and good friend Sam said, "We sound like machines!" And we did; suddenly, we sounded like perfect robots hitting every note with exacting precision we had not yet actually come close to achieving. The players we were listening to were us, but at the same time not us. Our performances had been edited so that we were much tighter, albeit possibly less…human versions of ourselves. Coming of age in the heyday of grunge rock we weren’t even sure if precision was cool (still a question I struggle with). However, after hearing the song several times in its new brutally mechanistic incarnation, we began to internalize the more precise rhythms. The next performances of the song would grow tighter and tighter until we in our human bodies achieved a nearly machine-like proficiency that more closely represented the performance assembled “in the box” (in the computer). We became more machine-like, or perhaps just less rhythmically flawed, after a human-controlled machine showed us the way.
This symbiotic relationship between human beings and their tools is not new. This is the way it has always been for our species, as early as cave painting. Tools change us as we change our tools, and this is an integral part of learning. Computers are tools, just as the pencil is, and our relationship with our tools has given rise to such wonderful feats of creative achievement from the Yungang Caves of the northern Wei Dynasty to Beethoven’s 5th to whatever Hollywood blockbuster has most recently blown your mind (sand worm riding in Dune 2, anyone?).
Of course there is now little art that is untouched by computers. Most painters I know are creating reference images with digital tools or using apps to check the proportions of their drawings, or painting in large part in order to share images of paintings to social media platforms such as Instagram. There’s a kind of reflexive digital loop, and sometimes it feels as if we humans are simply cranking the wheel. This is an explicit part of the content of my own creative work. Tools and more tools. Most of the time, these loops and exchanges are productive and inspiring. Art and music today are more diverse and exciting than at any other point in human history. We are simply awash with ideas and sounds and images and yes, overwhelmed by the sheer quantity and diversity of human ingenuity.
Naturally, I do feel fortunate to have lived through a time when tools were a bit slower and more tactile. I learned to draw from books I checked out at the library. And I learned to play the guitar from other humans in the 90’s. Specifically, I learned to play from my good friends T.J. Miller and Travis Delaney during lazy afternoons in our dimly lit suburban bedrooms. I would simply watch their hands and put my hands on the neck of the guitar in the same way they did. Occasionally I would ask them questions. Then, I’d go off on my own and see what else the instrument could do. Viola! I got pretty good. We all did. We got a record deal. I never had to take a lesson. There were no apps.
Although generative AI promises to upend and confuse artists’ contemporary relationship with technology, I would argue that much art and music in the last thirty years was already informed by computer technologies. Just as recording artists create and modify arrangements and performances “in the box”, visual artists now make use of Adobe Photoshop, Illustrator, and other digital image manipulation software to generate imagery to then paint or draw from, print, 3D print, etc. And just as musicians have learned from computers, artists too now borrow from the aesthetics of computer-generated imagery. Even the physical, tactile products and performances of our age are heavily informed by digital aesthetics. And somewhat paradoxically, even most sepia toned images or retro-tinged recordings have been rendered that way by image or audio software plug-ins. The vinyl records that millennials suddenly began purchasing contain songs that were all created with the aid of Mac and Apple computers, just like everything else. We cannot escape the digital age; we can only pretend to.
In my own recent creative work I am treating AI just like any other tool. I am using it to disrupt and inform images and sounds and ideas that I already have. And from there, learning and “processing” new approaches. In the same way that Photoshop and Logic (sound recording software) have informed the way I create images and music, the aesthetics of AI are already creeping into my work and those of my peers. As a creator and a digital native, these tools are irresistibly interesting and efficient and the dramatic adoption of these tools and flow of capital is evidence that like me, many creative people have already incorporated AI into their daily lives.
However, while AI generation might be good at enhancing or augmenting an idea here or there, I am still firmly in the driver’s seat. As artists, this is not only our responsibility it is our calling, our desire. Without creative agency there is no reason to create, and no serious artist would willingly forfeit their creative vision to machines. For the layperson or hobbyist, perhaps generating something with AI is enough. You certainly can, and then you can call it art. And who am I to argue with what gets called art? But for artists that have long been engaged with “generating” their own work through non-AI assisted means these tools are still just tools. The software augment, speed up, creatively disrupt, or otherwise enhance our vision or more easily allow us to achieve it. Like many other people, I do believe that AI promises to unlock greater human potential at least in the short term and will probably change the aesthetics of music and visual art just as other computer tools have changed the art we have today.
The threats that AI poses are not about creativity. The threats are about rights, royalties and exploitation.
The ever-shrinking rights of creators in the digital age
We have seen this story before. Just as the advent of Napster and Instagram essentially rendered songs and images worthless to creators, AI is poised to wring each last remaining ounce of value out of the products of human creators and hand it to tech entrepreneurs. The 21st century is so far a story about transferring value and wealth away from human creators and thinkers to the owners and maintainers of siren servers (Jaron Lanier’s term) and digital infrastructure. The value is no longer in the now-ephemeral, zeros and ones that represent creative products themselves but in the storage and transference of those works. Storage and transference are arenas that individual artists and consumers cannot readily or efficiently access or extract value from. Human created data is the resource of the 21st century and the wildly aggressive search for new data to run increasingly powerful AI tools is evidence of this.
The only 20th century analogue that exists for the idea of protecting the value of transference of works for creators is performance royalties. To me, art is of little value until it passes from the artist (or machine) to the audience. This is reflected in the law: performance royalties are fees paid to songwriters and publishers when their music is broadcast in public. Performance royalties were and are a remarkable feat of copyright law, ensuring that songwriters are paid not just for sales of physical CD’s or records but for each time that song is played on the radio or TV etc. Regrettably, with the unprecedented personalization and atomization of entertainment experiences controlled by streaming services, performance royalties are not as lucrative as they once were. Additionally, to my knowledge no such performance protections ever existed for images (this is my own speculative idea that the creator of an image would get paid whenever that image appeared in public or on other devices).
To conclude on something of a a down note, I do not have great faith that artists or even record labels are well-positioned to stem the tide of AI capital, nor that congress has any understanding or power to quickly create and enforce laws that will actually protect artists. The work of artists has been floating around in the digital ether for decades. It’s there and we all put it there and we continue to release our work and ideas essentially for free on these platforms every day. This is the world that silicon valley built, and we fell into it like a bottomless well.
Last week, a group of well-known (and lesser known) recording artists released a statement against “irresponsible AI”. This statement is vague and offers no solutions or plans, only that tech companies should be “responsible”. This strikes me as the same kind of short-sighted hand-wringing that the music industry employed in the early aughts. As record labels in the aughts were suing children for amassing large Napster mp3 collections on their personal computers, young “tech innovators” were assembling the code that would become Facebook, Instagram, Spotify and the like. Corporations are rarely swayed by suggestions that they simply try and be responsible (incidentally this suggestion does not work on my four-year-old son either!). Rather, we must change the laws and change them now to create a performance rights system that would reward creators when their works are used for training data or when works are “performed” on screens in any kind of format or capacity.
Some C words come to mind: consent, compensation, control and credit. These are the things that artists should be defending. Does this company have my consent to use my image? Will I be compensated if my images or songs are used as training data for AI tools? Can I control the use of those songs or that material? Will I get credit? Sadly, the answers to all of these questions are already no. Given the already vast chasm between the power of silicon valley and the power of creators, it is hard to imagine that artists will be able to do anything but play defense, and play it badly, for the foreseeable future.
In the meantime, as the old copyright laws are rendered useless we can lock ourselves in our studios and enjoy the process of communing, exchanging, creating with computers. What else would we do?