Studio Ghibli AI trend sends us back to simpler times

By Gearoid Reidy

For a brief moment, everything was Ghibli.

We all know that social media algorithms elevate material that you’ll engage with — typically content designed to enrage. But one instant last week felt like a reprieve when, following the release of a new OpenAI image-creation tool with few guardrails, suddenly everything was decked out in the telltale style of the Japanese animation studio.

Why it went viral is anyone’s guess. It seems to have begun with an innocuous tweet noting the brownie points attainable by converting family photos into Ghibli style. GPT-4o can of course do other styles; it could just as easily have been shots redone in the style of One Piece, the Muppets, or Rick and Morty.

But the people demanded Ghibli and, within hours, everything that could be Ghiblified had been. Perhaps the reason it took off is the sheer cosyness of its world, particularly amid a frantic global environment where old friendships are falling apart, and a rapidly changing economy with AI at its fore. The Ghibli images add a layer on reality that renders the mundane magical. In most of the studio’s movies, even the villains are understandable, and who among us has not wanted to ride the Catbus from My Neighbor Totoro, or seek comfort in the onigiri rice balls from Spirited Away.

Another reason it exploded is surely the juxtaposition of Ghibli’s dreamlike world onto real-life horrors. Some of the most darkly funny and widely shared uses of the filter were the most tasteless, from a Ghibli JFK assassination scene to the Ghibli Stalin in the infamous photo from which the purged Nikolai Yezhov was removed.

Of course, it didn’t last long. Such trends are only interesting so long as they remain organic: It was bad enough when OpenAI Chief Executive Officer Sam Altman changed his profile picture to a cartoon version, but once the White House account tweeted a picture of a Ghibli President Donald Trump slapping handcuffs on a weeping fentanyl dealer the trend had already gone from whimsical to cringe, in less than 48 hours.

Hot on its heels was a backlash from artists. Complaints ran the gamut, from legitimate concerns about future employment prospects and complaints about the wealthy using the tool instead of commissioning a human artist, to a head-scratching theory that AI art itself is inherently fascist.

Many noted that Hayao Miyazaki, the genius responsible for the studio’s most iconic work, would not approve. Yet his well-known quote on AI art, in which he seems to declare it “an insult to life itself,” is often taken out of context. Miyazaki was speaking specifically about a demonstration showing AI animating creepy, zombie-like characters with missing limbs, which he felt insulted disabled people. In the same documentary, however, he does mutter that the demonstration made him feel that “we are nearing the earth’s last day.” While he hasn’t publicly commented on the recent trend (assuming he is even aware of it), it’s a fair supposition that he wouldn’t approve, at least when it comes to the imagery being used by the White House.

Yet the knee-jerk reaction that conflates all AI work with slop seems overdone. Don’t get me wrong: I despise the grifting bros who two years ago were flogging ape NFTs; the fake accounts that mindlessly regurgitate artificially generated drivel, and the engagement farmers that abuse it.

But like any other tool, AI art can be used for good or ill. For every tasteless White House tweet there was someone getting joy from the technology. Even if Miyazaki is no fan, he did integrate computer animation tools into his studio after a long period of opposition. Studio Ghibli even released a fully 3D-animated movie in 2020, directed by Miyazaki’s son Goro, though it looks like a bad ‘90s videogame cutscene and was received about as well. Nonetheless, computer animation itself greatly expands our ability to turn vision into reality, and studios like Pixar have shown how to imbue it with soul — which is the thing currently missing from the worst attempts at art.

There are lots of gray areas, but I hesitate to jump to conclusions. Art style can’t generally be copyrighted — ChatGPT will make almost anything Ghibli-ish, but it won’t draw Totoro, at least not without some workarounds. Japan’s laws around AI scraping are tremendously permissive, as my colleague Catherine Thorbecke has noted. Perhaps the creators may bring some legal cases, but if anything this whole episode has been a free advertisement, demonstrating the strength of their brand. The online Ghiblis will fade, if they haven’t already, but as the critics note, the studio’s work carries far more depth than what amounts to little more than an advanced Snapchat filter.

In any case, this tech isn’t going back in the box. AI art certainly has implications for artists and animators, just as automation has in the past come for bank tellers or travel agents, typists and toll-booth operators. That these jobs aren’t creative doesn’t mean they don’t matter, and likewise creative work is not a uniquely protected category. Japan is suffering from a crippling lack of animators, who work long hours for little pay. If AI tools can help alleviate that, it should be welcomed.

This pace of technological change is another thing that rattles our brains. And ultimately it’s precisely because the world is so full of such uncertainty that we looked to Ghibli — to briefly let us inhabit a simpler, more comforting one.