AI is not "democratizing creativity." It's doing the opposite
Why Silicon Valley's favorite AI buzz phrase is so misleading and insulting
Greetings, and welcome to another edition of BLOOD IN THE MACHINE, a newsletter about big tech, AI, labor, and power. This newsletter is free to read, so please sign up below, thx, but it’s made possible by those of you who pay to subscribe. It means a great deal, and makes the continuation of this work possible. Thank you. Onwards, and keep those hammers at the ready.
You’ve probably heard the news: AI is going to “democratize creativity”. It’s also going to democratize medicine. And education, design, innovation, and, somehow, even knowledge itself. Few AI buzzphrases have stoked my anger as much as this one1, given that AI companies, of course, are in fact doing something closer to the opposite—giving management a tool to attempt the automation of jobs and execs a chance to concentrate wealth while promising benefits for all. And it’s everywhere.
When Arianna Huffington announced the launch of a new startup, Thrive Health AI, backed by her company, Thrive Global, and OpenAI, she wrote that the “company’s mission is to use AI to democratize access to expert-level health coaching to improve health outcomes.”
When OpenAI Chief Technology Officer Mira Murati caused a(nother) wave of backlash for saying that her company’s automation software was going to kill some creative jobs but “maybe they shouldn’t have been there in the first place,” she was quick to deploy the “d” word in her mea culpa:
“Moving forward, I believe AI has the potential to democratize creativity on an unprecedented scale. A person’s creative potential should not be limited by their access to resources, education, or industry connections. AI tools could lower the barriers and allow anyone with an idea to create.”
Now, this language long predates the current AI boom; as long as I’ve been covering the industry, tech companies have been including the dubious verb in their pitch decks and press releases. RobinHood pledged to democratize finance. Uber was going to democratize transportation. Maybe most famously, Theranos promised to democratize medical testing, or all of healthcare, depending on the day.
I think it’s mostly been understood, or at least intuited, that “democratize” has typically amounted to ignorable Silicon Valley jargon, ubiquitous but generally meaningless; the “4 out of 5 dentists recommend Colgate” of promoting your startup. Something that a founder or company is expected to say, because all their competitors say it, to assure a prospective investor or partner or tech media outlet that it is aiming for as broad a market as possible—but something that few outside of the industry would assign any real credence or even pay attention to.
“We want to democratize x” essentially translated into “we want to get as many customers to use this product as possible”, which, in its belabored effort to layer an altruistic sheen onto a corporate pitch, may have been eye-roll inducing, but it was noncontroversial enough.
Unfortunately, this has shifted: While AI companies are certainly using the term to pitch their products as having broad appeal, they’re also leaning on the notion of “democratization” as a way of countering growing concerns over the damage those products stand to do to creatives. They’re trying to get the term to do real work. And “AI will democratize creativity” is perhaps the worst offender. Specifically, there’s this frankly absurd argument that much of the AI industry seems to have embraced—best exemplified by the above Murati quote, though she’s far, far from alone in advancing it—that AI is somehow leveling the playing field for would-be artists. That AI products will break down the barriers erected by the art world’s gatekeepers, and allow anyone and everyone to finally be creative.
The first time I heard the “democratize creativity” tagline was at a Copilot demo event in LA, where Microsoft was presenting the AI software to influencers. The speaker showed how Copilot, powered by OpenAI’s GPT tech, could create images with the simple inputting of key terms into a text field. “I never thought of myself as creative,” the Microsoft rep said, “but it turns out I can be.” (He just needed some software automation, it turns out.) More recently, Justine Moore, a partner at the VC firm Andreessen Horowitz, which has invested heavily in AI, has been making a similar case as Murati to marshal support for the embattled technology. In a thread that concluded that “the panic around AI art is overblown,” she insisted that “all AI creators are artists.” A few months ago, I had an exchange with a member of Microsoft’s AI team, who countered my assertion that AI could erode working conditions for creative workers by asking, ‘well don’t you think it democratizes creativity, too?’
But the ‘AI democratizes creativity’ line finally became unignorable when Murati, who helps run the most influential consumer AI firm in existence, issued her damage control statement. Now, I think part of this defaulting to the “d” term is that OpenAI and others have been caught flat-footed by all the protests to their products by artists, writers, and creative workers—I do not think that if they had properly anticipated such a widespread backlash against AI, they would choose “it uh democratizes creativity” as their primary PR defense against the fact that it threatens people’s livelihoods.
This is because if you spend more than 45 seconds thinking about it, rather than allowing it to glitch past at 2x speed on a business podcast, it becomes so patently ridiculous that it is almost offensive. I know it is offensive to many artists, especially since it’s deployed in service of achieving *precisely the opposite aim* that it purports to. AI will not democratize creativity, it will let corporations squeeze creative labor, capitalize on the works that creatives have already made, and send any profits upstream, to Silicon Valley tech companies where power and influence will concentrate in an ever-smaller number of hands. The artists, of course, get 0 opportunities for meaningful consent or input into any of the above events. Please tell me with a straight face how this can be described as a democratic process.
The other thing that really irks artists and creatives is that making art is already a fundamentally democratic process. Anyone can do it! (Hence the pick up a pencil meme.) It just takes time, effort, training, dedication, a development of craft. AI advocates have tried to argue that AI helps disabled people create art—but the already plenty vibrant disabled artist community shut that down extremely quickly. No, it’s making a living practicing art that is the tricky part, the already deeply precarious part—and it’s that part to which the AI companies are taking a battering ram.
It’s true that, as Murati points out, not everyone has the right industry contacts, but how does AI change the equation there? Besides, that is, making matters worse? With AI giving rise to a flood of samey-looking AI output, if anything, industry connections only matter more; the science fiction magazine Clarkesworld had to close its submissions, as its editors no longer had time to wade through reams of mediocre ChatGPT output, and turned to working only with writers they recognized or already had relationships with. As far as I’ve seen, no one who’s arguing that AI is a harbinger for a new democratized paradigm of creativity has offered an explanation of how the current gatekeepers might be done away with, what that might mean for a society with a functioning creative economy, or how the industries that creatives rely on to pay rent will in any way be made more equitable by its arrival.
Of course they haven’t. To the big AI companies, none of that enters into the equation. The democratization pitch is aimed not at aspiring artists, but at tech enthusiasts who may or may not feel that largely abstracted gatekeepers have been unkind to them or derided their cultural contributions, who feel satisfaction at seeing slick-looking images produced from their prompting and eagerly share and promote the results, and industries who read the ‘democratize’ lingo as code for ‘cheap’, and would like to automate the production of images, text, or video.
The AI companies, of course, do not much care if they take a wrecking ball to the already fragile creative economy. Creatives are already losing work, seeing pay rates decline, and are being asked to use AI to improve their productivity. In another of her defenses of AI art, Andreessen’s Moore argues that AI helps artists produce more output—which, hooray! Artists get to crank out more work for the same or, soon, lower price, spending less time on craft and more on jamming the generate button, and so the value of art on the market tumbles down. People often say they wish the tech set would take more humanities classes—I wish they’d study some political economy.
And that’s just the “democratization of creativity.” I’m not even going to wade into the intellectual bankruptcy of the idea that AI will democratize healthcare or education here, in part because those notions are somehow even more insultingly and obviously opportunistic. Education and healthcare face serious, structural problems, and the idea that they could be “solved” or even meaningfully ameliorated with chatbots who make mistakes 20% of the time should immediately strike any reasonably intelligent observer as unserious. (This is not to say that there aren’t use cases for AI in either field, but the idea that it will “democratize” sectors in desperate need of reforms to make more affordable, or funding for adequate supplies and pay, is preposterous.)
No, AI is not going to democratize a whole lot, I’m afraid, aside from things like ‘access to second-rate customer service bots for midsized business owners’. It is, by and large, a profoundly anti-democratic force, in fact, given that who decides to adopt it in a workplace is almost always management, that rank and file workers are almost never given any input into whether or how they might want to use it, and freelancers must simply deal with the economic fallout of declining wages and fewer opportunities.2 Again, this doesn’t mean there are not specific utilities for which AI might be useful; but describing the act of generating automated output as ‘democratizing’ is nonsensical at best, and insulting at worst.
There’s plenty that companies like OpenAI could do if they were earnestly interested in “democratizing” anything. To start, they could compensate writers, artists, coders, and other creative workers for the material they’re already training their automation systems on. They could seek meaningful consent from these workers as to how or whether they’d like their stuff treated in the training corpuses. But they won’t. Because the major AI companies aren’t interested in democratizing creativity—they’re interested in transmuting it into profit.
There IS another one, and I’ll get to it in due time…
It’s not lost on me that AI companies are making their democratization pitches, which inherently rely on a pointed dumbing down of the very concept of ‘democracy’, just as we have entered a period of unparalleled handwringing over whether the real thing is going to make it through the next presidential administration intact.
"Democratize" is simply Silicon Valley's misspelling of "displace".
As the wife a saxophonist, I have learned that, "democratizing creativity" means redistributing more of what's left of the wealth to the wealthy. Long before AI chatbots arrived, residuals (pay for musicians) were gutted in the US in deals struck with music platforms (I'm talking about you, Spotify), while outside the US, their artistic contributions remain compensated. Today, AI enables reconstituted copy-cat theft at scale. (Just name your favorite artists and ask DALL-e to create an image.) I doubt laws can contain it because this easily reaches far beyond US borders making it difficult to stop.