Thieves in the Temple
AI is here and, as someone who does stuff on the internet for a living, it's annoying me on a daily basis.
The content scraping revolution has taken my workplace by storm. As disruptors at the forefront of cutting-edge innovation or something, we now use a semi-functional chatbot to do our work for us. When the servers are up, anyway. Client-facing messaging? Done. SEO content? Covered. Ad copy? Locked in. We're so productive, using bits of information and language scraped from the depths of Google and reassembled to just barely pass a freshman core class plagiarism checker. The future is bright, baby.
Part of my job these days is learning how to make creative assets using Midjourney. Boy, do I hate it. Teaching a machine to dream up stock images takes more time and effort than just putting something together in Photoshop myself. Remember, friends, we're efficient now. AI is such a gift. One day, I sat at my work computer and watched everyone else in one of the Newbie servers work and rework their prompts. I was there attempting to wrangle Midjourney into conjuring images I could find in a stock repository in ten minutes. There, watching the images roll by, I saw someone try to make a photorealistic fanart of Giorno Giovanna from JoJo's Bizarre Adventure.
Obviously, I stopped whatever else I was doing to watch this guy make a photorealistic fanart of Giorno Giovanna from JoJo's Bizarre Adventure.
I am vaguely aware of the processes involved in creating images with AI. I would never call these images "art." While I can't adequately describe it to someone else if asked, I've read and seen enough on it to loosely grasp it. I always found the experience of "creating" with AI to be frustrating and arduous. Whatever I've eventually gotten the AI to burp up always felt cold. Hollow. There's an emptiness to it, a death behind the eyes of the rendered figures or in the sunshine streaming through an approximation of windows that no amount of tweaking can overcome. But as I'm sitting there watching this person make Giorno fanart, I find myself fascinated by the process.
This, too, is just scraping. I can see what images Midjourney is taking from with each step of the process. Beginning with Hirohiko Araki's lineart of Giorno from Part 5: Vento Aureo and moving toward Hugh Dancy's spread for Burberry's Autumn/Winter 2004/2005 men's collection.
Nope, Hugh Dancy is too old. Red carpet photos of Timothée Chalamet next, Getty watermark and all. The Met Gala, I think? Maybe the Oscars?
Nope, the hair is wrong. Try again. Try again. Try again.
There's something kind of wholesome about it, at first. Someone, somewhere, really wants this image of Giorno Giovanna. Working, reworking, tweaking and tweaking. Getting it just right, to match some image they have in their head, the platonic ideal of photorealistic Giorno Giovanna fanart. Who knows what they even want it for. To save? To sell? To show to their friends? To use as their phone wallpaper? Sitting there, watching each iteration of Giorno be born from the ashes of reworked inputs, it didn't really feel any different than drawing Giorno from a celebrity photo reference.
But then I noticed the death behind the eyes. The uncanny stare of something made just a little…off. The way the pupils don't look right. The way the eyes look the wrong kind of wet. This isn't Giorno Giovanna. This isn't Hugh or Timothée or any other smooth, svelte, little cherubim of a man used as a seed image. This is something…else.
I close the tab on my work computer.
I haven't used Midjourney since. I'm fine with that.
Love, come quick, Love, come in a hurry
Because the AI revolution is here and I can't sprint fast enough to escape its teeth, I've been trying to wrap my head around the concept of AI-generated art. By art, I mean portraits and poetry, books and landscapes, character designs and game scripts. Imaginary things made real, or some facsimile of it. I've been reading saber-rattling articles about the disruptive force of AI in academia. Cheerful business website listacles launder techbro dystopia with the promise of increased productivity. Everyone raises points. I find all of the points a bit lacking.
If not lacking, then just…falling short of helping me make sense of this in theory or practice. There's also the chance that I'm conflating a lot of different discussions and ideas because this is above my pay grade. Which is fine. It's a good thing I'm just a dipshit with a newsletter and not a thought leader.
Some people are fairly sunny on the application of artificial intelligence and its role in the arts. AI-assisted writing tools already streamline certain creative processes for writers. They predict your next word, tidy up your grammar mistakes, and can help you brainstorm by generating ideas when you're stuck. I hear it can be a bit comforting to ask a chatbot for help if you're suffering from writer's block. To get quick advice on the fly when someone else may not be there to listen, and to jot down its answers and let your brain turn over until it finds a solution. After all, that's all we're talking about here. Just overcoming stumbling blocks. Tweaking. Fine-tuning.
According to writer and AI-human interaction researcher Katy Ilonka Gero for Wired:
It’s easy to be for or against including computer-generated text in your work, but I predict the conversation will get much more nuanced as we encounter the various ways computers can impact our writing. Writers want to protect their authenticity and intention. It may be useful to think of computer-generated text as dancing with the writer’s text. Most writers aren’t against collaboration. They just want to be in control of their own dance. As long as the computer can match the writer’s footwork, writers are happy to let a computer contribute to the performance.
But people have been experimenting with AI-generated fiction for years. I think of The Castle Freak, the remote writing residency program for generative composition from Inside the Castle. Each year, a writer "writes" exactly 100,000 words in five days without violating copyright laws. Whatever comes out of that is not human and was not created by a human consciousness. It was programmed and compiled by human hands but is not, by definition, a human work.
The mission grounds itself on the anecdotal precedent set by Stephen King in On Writing:
At the end of my adventures I was drinking a case of sixteen-ounce tallboys a night, and there’s one novel, Cujo, that I barely remember writing at all… I like that book. I wish I could remember enjoying the good parts as I put them down on the page.
As implied by King and explored by the program, "A human consciousness did not write Cujo." Joanna Walsh's 2021 book, Autobiology, works within this framework to follow the central idea to a fascinating conclusion. It is a book written by an AI trained by Walsh. I have not yet read it in its looping, winding, dreamlike entirety, but it is fascinating to take in.
Now we have machines that we can train to write books. Sort of. We already have them writing SEO articles and trying to game the Google algorithm for better search rankings. That's easy. But a book? I assume a book produced by a well-trained, carefully watched ChatGPT would be more aesthetically pleasing and narratively coherent than Walsh's inhuman offering. One day, with time, a lovingly automated piece of fiction would probably feel close enough to human as to pass the Turing Test.
But would it be satisfying?
Writing for The Spectator, Sean Thomas reads doom in our forecast:
Putting on my pointy hat of pessimism, here’s how I think it will pan out. The machines will come for much academic work first – essays, PhDs, boring scholarly texts (unsurprisingly it can churn these out right now). Fanfic is instantly doomed, as are self-published novels. Next will be low-level journalism, copywriting, marketing, legalese, tech writing; then high-level journalism will go, along with genre fiction, history, biography, screenplays, TV drama, drama, until eventually a computer will be able to write something like Ulysses, only better. The only prompt will be ‘write a long amazing novel on whatever’.
The argument that I keep seeing is that writing is important and valuable. It is a skill that must be trained and sharpened like a knife. Writing in school and at university teaches us how to assemble our thoughts and construct arguments, discovering what we really believe and value as our fingers fly across the keys. Our embodied experiences leave stains on the page, wine-dark on toothy paper despite their ephemerality. Things that can't be touched but can be known. That spark, flesh-like and electric, that tells you that this is a human effort.
People assure me that we need that touch to feel. They tell me that ChatGPT will simply generate keyword sludge until all the bots realize they're being fooled and the AI content bubble will burst in a sweeping algorithmic shift. The bots chattering online can be fooled for longer but we, the warm and breathing, will know immediately. We will always want a human heart, they say. We can't be fooled.
But I don't know.
The profit motive is there. Who needs a ghost writer when you have a machine? Who needs to write 50,000 words a month, every month or so, to feed the Amazon algorithm if you can spit out a book a week? Every month can be National Novel Writing Month, forever. Produce a completed comic book without an artist or a script. Write a game and design all the characters without having to pay anyone. How can anyone compete with that?
We're already content, you and I. Artists and writers and journalists and activists and influencers and experts and educators. We chase the trends and pivot to whatever platform or format is trending. Perched at the ready, moving as quickly as our soft, failing bodies will allow whenever the algorithm barks at us. Search engine optimized lives. I mean, we have Frankensteined pop stars these days, the dead revived by holograms. Long gone actors reprise beloved roles and singers release new music from robbed graves.
If you told me the next Disney film was written by AI, I would believe it. I wouldn't be shocked if the next show I watched on Netflix was written by a chat bot. Dead actors reading machine-generated lines, their stolen voices remixed and regurgitated. Tell me you have tickets for Prince next week and I wouldn't bat an eye.
Even if it looks and feels dead, are we still alive enough to care?
Love, if you're there, come save me
So, poet and writer Joseph Fasano posted this to Twitter recently.
I'm not familiar enough with poetry to speak on this with any authority, but I assume it is quite good because I've been thinking about that last line for a while now. What is love in the age of content? I'm not sure yet.
Feel like I'm looking for my soul
So, I've been thinking: Weren't synthesizers supposed to do away with instrumentation?
That's what I wanted to know as I started reading up on AI art. The concept of "electronic music" as we probably think of it today dates back to 1876 with Elisha Grey’s Musical Telegraph. A kind of tiny keyboard, the Musical Telegraph was built to create and transmit musical notes over telephone lines. It sounds downright adorable to think about now. Electronic music was expanded upon and popularized in the collective imagination through the 1940s by a range of musical inventions and experiments, such as Thaddeus Cahill’s Telharmonium, Lev Termin’s theremin, and Lauren Hammond’s electronic organ.
In 1946, composer, musician, and engineer Raymond Scott founded his company Manhattan Research, which eventually created the Electronium. The Electronium was an “instantaneous composition-performance machine” that composed and produced music based on user inputs without keys, strings, or other human-instrument interface. Scott's machine was designed specifically to be guided, programmed, rather than played. Musicians were not pleased by this idea, nor Scott's drive to remove human interference -- the spirit, the essence -- from the process of music-making.
Though it was never fully completed and mass produced as a commercial product, this machine shared much of the same DNA, as it were, with the world's first true synthesizer, the RCA Mark II Sound Synthesizer. In an essay for Science Fiction Studies, Nicholas C. Laudadio writes:
Yet not surprisingly, the RCA project and the Electronium’s attempt to remove “the human element” from musical creation represented for many the most dangerous consequence of the technologization of musical production, threatening to replace live musicians and turn music into a purely industrial commodity. These infernal devices, with their Fordist air of automated manufacture, stoked fears that machines in music could (and would) lead to what some saw as “the ultimate technological nightmare.”
While I understand that synthesizers and chatbots are very different things, it's interesting to think about. These things just keep happening, the supposed automation of creative processes. As I was looking into the history of synthesizers, I found an article in The Oklahoman from 1990. It's an interview with a sales manager at a local music store, a man named Darren Tackett. Tackett remarked on the power of electronic music. I thought it was very nice. Very quaint.
"For the longest time, music stores sold mainly to musicians. But now, with the advent of electronic instruments, people with no musical training can enjoy the music-making process."
"You can use IBM, MacIntosh, Atari and most other personal home computer systems. There are computer programs for the non-musician who has dreamed of playing something, but wasn't able to take lessons."
…
"Technology does not replace musicianship, but it does give it more tools."
I bring it up at all because I like electronic music. I listen to lovingly chopped and screwed arrangements made by people living in studio apartments, who work day jobs like I do. Their music is made on machines, but it still feels human. It certainly looks human when I watch my favorite producers at a live set. luxury elite is real and I love her. (Respectfully and platonically, of course.) The tools she uses feel human in her hands. I mean, hell, Prince used to program flaws into his drum machines so they sounded more human.
The tools are useful, too. I use ProWritingAid to spot errors in my writing, or to double check something I've edited to make sure I didn't miss an extra space or some errant punctuation. It checks to make sure the prose sounds organic. Natural. We all use AI translation tools to casually check words or sentences from languages we don't understand. Predictive text makes sending texts easier when my hands or joints hurt. Scrolling online, I find posts about people with dyslexia or language barriers using ChatGPT to draft emails and messages. AI is a tool the same way the synthesizer is a tool.
I won't buy AI art, though. I don't feel comfortable making it, either. Maybe it's a stupid distinction. I don't think there's anything particularly reprehensible about someone harvesting surreal, abstract, or distorted images from the stream of a machine's weird, worm-like unconsciousness. Horror and surreal art is right at home alongside the too many teeth and multitudinous fingers of AI-generated compositions as it stands. The uncanny nature of machine-produced art, like King autonomously writing Cujo while black-out drunk, makes sense to me in that context. Walsh's generative book makes sense to me. But I am creeped out by the undeath contained within the photorealistic Giorno Giovanna. That's the distinction that I have to make for myself, whether it makes sense to anyone else.
Me and you could've been a work of art
I asked before if we're still alive enough to reject AI-generated art. The answer so far has been yes. As I was writing this, Neil Clarke raised grave concerns about the growing number of machine-plagiarized or -generated submissions Clarkesworld has received. Clarke had been tracking the number of these submissions, seeing the occasional spike. Any writers caught in the act were banned from future submission. After ChatGPT gained international attention in late 2022, the number of these submissions increased dramatically.
Now, Clarkesworld has announced that they have closed submissions due to the avalanche of AI-generated stories.
It may just be ChatGPT slop today, but it's slop clogging submission queues and inboxes. It's a slop that wastes the time of editors and slush readers such that they can no longer keep up with the rush of chatbot sputum. In the midst of an already turbulent publishing landscape, AI is now making it that much more difficult for writers to find publishers for their work.
At time of writing, Amazon has over 200 new books with the author or co-author listed as ChatGPT. They appear to be mostly simplistic children's books, sterile how-to guides, and lifeless poetry collections. This won't stop anytime soon.
It's all trash today, but as AI gets better at assembling and compiling narrative fiction, what do we do about these stories? How can we contain them? Now that work is being stolen, or repurposed, or reused, or however AI advocates want to call it, what do we do? There has been seismic backlash from every industry and creative medium regarding the use of AI to generate character art, book covers, and especially books and comics. But is condemnation enough? A man named Ammaar Reshi recently went viral (in the bad way) for producing a children's book called Alice and Sparkle in ChatGPT and Midjourney. Kris Kashtanova's Midjourney-made comic book, Zarya of the Dawn, was met with controversy and condemnation. The nature of authorship has called their copyright registration with the U.S. Copyright Office into question. At time of writing, the same office is involved in a lawsuit over denying copyright to works with artificial intelligence or software listed as the author.
I am neither smart nor well-read enough to discuss the complex legalities of the issue. Then we would get bogged down in intellectual property laws, fair use, and copyright. What is theft and what is fair use befuddles me. We end up in circular arguments about the innumerable atrocities of capitalism and the evils of owning ideas and property. Because if we just had a revolution and universal basic income, it would all be solved. But I'm not the person you need wading into that. I want working artists to have their work protected so that people can't just recreate their style for free and avoid paying them a fair wage. I don't want writers to compete with chatbots in a shrinking industry. I just don't know what to do about that in the situation we're in now, or the situations we're faced with in the coming years.
However, what I do know is that ChatGPT's creator, OpenAI, outsourced its data filtering processes to a company called Sama. Sama claims to be an "ethical AI company" and hires workers in Kenya, Uganda, and India to label data for American tech giants. Employees in Kenya making less than $2 an hour sifted through incomprehensible amounts of text and image descriptions so that OpenAI's cute chatbot wouldn't burp out Nazi propaganda and sexual exploitation material. Human beings were subjected to graphic descriptions of unrelenting violence and horrifying depravity for mere pennies, just so I can ask ChatGPT to write a rap in the style of Megan Thee Stallion. These workers are not alone in this, either. This is how it's done in the tech industry, off-loading moderation and filtering to people in the Global South. People we don't see who have to live with the darkest corners of the internet in their heads now because OpenAI and Meta didn't want to do it themselves.
In the most mercenary way, I understand that someone, somewhere, has to filter this data. The stuff we’re talking about is vile. But you want me to believe that it simply must be done under such grueling conditions without any support? That these tools simply must exist, must go to market as quickly as possible, and this is the only way they can?
As reported by Billy Perrigo in a recent expose for TIME Magazine:
One Sama worker tasked with reading and labeling text for OpenAI told TIME he suffered from recurring visions after reading a graphic description of [horrific things I will not repeat here]. “That was torture,” he said. “You will read a number of statements like that all through the week. By the time it gets to Friday, you are disturbed from thinking through that picture.” The work’s traumatic nature eventually led Sama to cancel all its work for OpenAI in February 2022, eight months earlier than planned.
This is what haunts me. I really want to highlight this next part here, because this is what pushed me to write the piece you're currently reading in the first place:
Three employees told TIME they were expected to read and label between 150 and 250 passages of text per nine-hour shift. Those snippets could range from around 100 words to well over 1,000. All of the four employees interviewed by TIME described being mentally scarred by the work. Although they were entitled to attend sessions with “wellness” counselors, all four said these sessions were unhelpful and rare due to high demands to be more productive at work. Two said they were only given the option to attend group sessions, and one said their requests to see counselors on a one-to-one basis instead were repeatedly denied by Sama management.
For all my whining about how bad ChatGPT and Midjourney are, how conflicted they make me feel, this is the cost. My productivity, my amusement, my ennui, are not more important than these workers' well-being.
The thing is, I get it. I understand how AI and machine learning can be used to benefit people. How chatbots can make communication easier for people living with disabilities or language barriers. How Midjourney can allow a hobbyist to create silly JoJo's Bizarre Adventure fanart that they couldn't otherwise make. I see people use AI to sketch out ideas for book covers or other materials that they later take to an artist or designer, just so they have something that makes visual sense for reference. Machines check my grammar. If you can't draw and use a tenderly worked prompt to make something beautiful out of it, because you just want something beautiful, I don't think I can be angry about that.
But the revolution we have been sold is coming on so hard and so fast. The limitations and complexities don't matter to those championing its integration into our lives. Artists have their work scraped without their knowledge or consent so machines can be trained to vomit their style back out. Illustrators and cartoonists are having their marks stolen just days after their deaths. Voice actors fear their performances will be taken from them and replaced by speaking machines. Everything is just a data set waiting to be exploited. A series of values. We're all told that it's progress, and there's no stopping it.
Tech companies sell us tools that may or may not exist, may or may not function, on the empty promise of innovation. Tools that, like everything else we enjoy, rely on exploiting and traumatizing people whose suffering we don't see. Your company is more valuable if you leverage the latest technology and slim down your workforce for maximum efficiency, so you make your organization adapt whether the tools work or not. Companies rely on Dall-E and Midjourney recreations to avoid paying for real art or photography. Words written by human hands are plagiarized, recycled, and sold as new, fresh SEO content. Buzzfeed laid off staff writers and replaced their work with ChatGPT. Other companies will follow suit. Progress, they tell us. People will lose their jobs. Their lives will be made worse. Progress continues marching on, dragging us behind it wherever it goes.
It will all be for nothing. Then it's on to the next revolution.
And so, I keep thinking about that poem by Fasano and its final line: Love is for the ones who love the work. We are all devalued. We are nothing but flesh but the flesh isn't good enough. Everything we touch bears the scars of capitalist expansion, the fruit withered on the vine, the flood's faded lines on the wall. We are broken. I am broken. Tired. It would be so much easier, so much faster, to off-load my craft onto a machine who can write for me. Maybe it could brute force my writer's block and ethics to finish the books that I can't. Maybe I would make more money that way.
But I love the work. The work keeps me alive. I have people on social media shrieking that by not giving myself to technology, to TikTok marketing funnels, to search engine optimized books, to search engine optimized identities, I am leaving money on the table. I'm not embodying happy. I should just be happy and make my little content. Be happy. It all sounds like the voice of a machine.
Make the content.
Be happy.
Be productive.
Be happy.
Embrace progress.
But I can't.
Art is all I have left as the water rises.
I pray that you can live for art with me.
Add a comment: