Generative AI Isn’t Worth It
A request to creative people
November ended with an anniversary. It was two years since OpenAI released ChatGPT to the public.
What a quaint time that was. Out of curiosity, naivety, late-pandemic boredom, or maybe some remaining shred of optimism about Silicon Valley, people took to it. Social media and group chats filled with machine-generated text that was at turns impressive, hallucinatory, funny, and terrifying.
The hype was huge. So was the concern, especially among the few people who still manage to make a living in creative fields. Will we be replaced? was the question of the day for writers, artists, and journalists. When, in early 2023, ChatGPT became the fastest-growing consumer app of all time, the word When was added to the front of the question.
This was no backlash, though, not in any sense that matters. Yes, people made fun of AI. They called the text and the images and the videos it produced “slop.” They worried over the future of democracy. Soon enough, many of the concerned creatives welcomed generative AI into their lives. People who I talked to about the machine’s creepiness sent me AI-generated emails. Some asked the machine to weigh in on outlines and research I sent them for projects. People who told me they didn’t trust AI were among the first to send a jokey AI-generated image or say “let’s ask ChatGPT” when our conversation landed on a question neither of us could answer. “AI is my writing prompt” is a phrase I’ve heard too many times. The cognitive dissonance has only grown in the last few months. Newsletters I subscribe to and BlueSky feeds I follow from self-proclaimed technology critics feature AI images. This season, a new meme has surfaced where people ask an AI to illustrate what it thinks their life looks like. Every image is some combination of computer hallucination and inside joke, rendered in AI’s insipid overly friendly, cliche style. As I write, I’m seeing BlueSky users switch from complaining about bad freelance rates to posting AI-generated roasts of themselves. The jokes write themselves—no need to ask a bot.
There are some useful AI tools. One person I work with uses an AI service that summarizes work calls. Another uses a bot to schedule meetings. I’ve talked to journalists who use AI to analyze data and help humans translate small pieces of text. In these cases, the machine is picking up busywork and not making anything except free time.1 I know there are uses in medicine and mathematics, too, which I’ll leave to be discussed by professionals in those fields. What I’m concerned with is generative AI—machines that write and make images—and the casual acceptance among creative professionals of technology that is meant to replace us.
This machine is not your helper. It is a thief. OpenAI admits that it cannot build its products without using copyrighted material. This explains the abundance of tired phrases in text and the blandness of the artwork. The technology is also terrible for the planet. Using AI to make silly memes (even memes about how bad AI is) is like rolling coal; destructive trolling that hurts you in the long run and makes someone else rich. But the most obvious reason to avoid using AI, at least for anyone who cares about creativity and quality, is that it produces garbage. Generative AI’s copy is unreliable and its images are an eyesore. It will get better, and it is getting better, but only with more stealing and more environmental degradation. Is that the price to pay to have a machine that you can bounce ideas off of or a funny-looking picture to post? Even if it is worth it for now, you’ll still end up paying with your job. This machine is already replacing artists and writers. Using it for any purpose is helping a business model that sees the elimination of creative labor as a path to profit.
So don’t. Don’t use generative AI. Do not give another inch to this technology. No curious clicks. No quick images to put atop a post. No get-a-load-of-this-garbage screenshot fodder. In two years of showing off their latest parlor trick, the venture-capital-backed plagiarists who profit from AI have done nothing to further the arts, nothing to strengthen the humanities, and nothing to improve any human endeavor that doesn’t make them personally wealthy, famous, or powerful. Don’t give them another user to exploit. Leave them to their sad lives of reinventing the city bus.
There’s a defeatism among creatives using generative AI. All the credulous news stories, the layoffs in the media industry, the dropping demand for work made by humans—it feels inevitable that this machine will be part of the future. Artists have seen their work devalued and trampled in a culture that idolizes the computer whiz, prays every day to data, and rewards whoever wields it with respect and money. But we should aim to remake this reality, not give in to trends. Even the helpful uses of generative AI should be unnecessary. A person should have time to write an email. They should have time to put thought into communicating. They shouldn’t have so many meetings that they can’t do the work they need or want to do. They shouldn’t be so overwhelmed with worries and busywork that they have to make a deal with the devil to meet deadlines. Time and money are already too tight to live the creative life we want to live. The machine seems like a solution. But if there is someone to blame for our lack of time and energy, it’s the same corporate class that venerates technology and doles out the meager pay. Saving yourself a little time now won’t make the future better, it just brings on your replacement all the faster.
Replacing artists and craftspeople has been a goal of technology for years. The first machines made manual labor less punishing, a true gift to humanity. The next wave turned to creative works. Machines turned crafts into processes and the owners of the machines turned those processes into products. The computer is a great tool for this because it works so quickly, and so invisibly, it may as well be magic. A 1961 issue of LIFE featured a computer that was meant to write “beatnik poetry” (with the smirking undertone that anyone could be abstract). In his 1966 book Most Notorious Victory, economist Ben B. Seligman wrote about computers that were programmed to churn out TV scripts. Now the machine is powerful enough that it can replace the last pieces of human craft with the algorithmic actions of a computer.
Technicians do not understand what artists and craftspeople do. By pushing their text and video generators as means of unlocking or enhancing creativity, AI firms imply that anyone can be an artist. But art has no value to the technician. That’s why the technician doesn’t see an artist’s job as something that’s been worked toward and earned. Instead, it’s something anyone can play at without needing practice. That’s why the technician steals the artist’s work to build its machine.
I mentioned the plagiarism problem to an engineer recently. He compared this theft to inspiration, as if the act of creating is little more than remixing, and as if everyone has the same finite (and limited) set of works that inspires them, and deploys this inspiration in understandable, predictable ways. He didn’t mean to be insulting; he truly seems to think that the real talent in making art is not the execution of an idea—of revising that idea through style, trial and error, and application of craft—but the fact of simply having an idea. In this view, anyone with an idea is entitled to have it realized, and it is technology’s job to make that possible with as little human effort as possible, no matter how much plagiarism and environmental destruction is required. How convenient that the ultimate control over how that idea is realized falls to the technician and his machine.
Behind the technician’s hubris is either a misunderstanding of human potential (they think a machine can make art because they think a machine can do anything a person can do) or a deep insecurity (they know that a decade of free money, lax regulations, and public adoration of the tech industry hasn’t led the industry to solve any of the world’s most pressing problems). Then again, maybe it really is just about making money. It’s certainly not about making art. This is not a tool of creativity built to sate the burning drive of an artist with something to say. OpenAI isn’t selling canvases to painters. Generative AI is a platform that concentrates power and money among a select few.
Maybe we can trust these companies again when they come up with a machine that does all of our chores so we can have some time to think.
If none of this resonates, then please, avoid generative AI out of sense of taste. Generative AI is an insult to anyone who likes to read, watch, or listen to art and media. The text and the images and the video from generative AI epitomize the phrase “good enough.” They look bad, with sloppy mistakes that the audience is expected to overlook because…well, the because is unclear. Because a machine made them? Because they were easy to generate?
AI’s makers expect their slop to win hearts and minds. But generative AI’s work is heartless and mindless. We shouldn’t settle for it. I don’t know if I believe in such a thing as a soul, but the times I’ve come closest to believing have been when I’m in the presence of a work of human creativity. It’s hard to make art or to hone a craft. That’s why not everyone does it, and it’s why so few people do it in public. But the attempt is the stuff of life. And falling short in the goal of self-expression is a reason to try again. If we don’t get up every day and try, why bother getting up? Trying to express ourselves with our own minds and our own limited abilities is like leaping over a chasm knowing full-well that death waits below.2 Using a machine to realize an idea is like gently crossing a bridge. Both may get us to the same end point, but which is the experience you’d want to hear about or see? ChatGPT is an Uber to the finish line of a marathon.
It’s either ironic or tragic that this turn toward machines is happening at a time of deepening distrust of our fellow humans, of outright hatred fueling political campaigns, of deep polarization, and of increasing isolation and loneliness. Art and creativity won’t solve these problems, but they can be reminders of our humanity. We shouldn’t run from that. Technology is alluring and authoritative, but that’s just another way to say it’s addictive. It draws us in with a suggestion that it can connect, inspire, and bring us closer. We end up feeling alienated, exhausted, and angry. Then we turn to technology again. It promises to solve problems that it only makes worse.
It didn’t create these problems. But then again, machines don’t create anything.
More from Together, Alone on AI:
With translation, I believe there must be “a human in the loop,” as they say. Language, tone, and intent are too nuanced and too subtle to leave entirely to a machine. I wrote about this application in journalism last year. I worry about the internet being flooded with “good enough” translations of news stories, which could only further deplete public trust in the press.
I’m borrowing this metaphor from the artist James Kochalka, whose two short essays Craft is the Enemy and Craft is Not a Friend shaped some of my thinking here. The essays, each written as letters, urge artists to stop worrying about polish in their art and to instead try to say something. I often hear people say that AI helps them make work that looks “polished.” I understand what they mean, but in practice, AI’s polish is simply an imitation of someone else’s work. Ultimately, this leads to sameness and it suppresses what might end up being someone’s own signature style. Maybe Kochalka’s best point about polish and style is his statement that some people think Spawn is well-drawn and Peanuts is poorly drawn. I know that someone could argue this from a technical standpoint, but as a statement, it makes absolutely no sense.