Nothing will ever be the same
The Ghiblification of Humanity
There were many important technical updates this week in the realm of AI — Gemini 2.5 is out and is actually quite good, two new open models have come out of China (Qwen2.5-VL-32B and the latest DeepSeek v3), OpenAI is introducing Anthropic’s MCP framework across their products, and Claude uncovered a Think technique. And yet, it’s almost as if none of that registered. For the last four straight days, the only launch on anyone’s mind (and timelines) was OpenAI’s new image model. But even putting aside the technical feat — incredibly impressive generation abilities, finally seeing well-done text rendering, and a new tech that’s not based on diffusion — really, what everyone was obsessed with was turning themselves, their loved ones, their dogs, and pretty much anything there ever was, into an image re-rendered in the style of Studio Ghibli.
I don’t know that I’ve ever seen that level of saturation, ever. Actually, the only other time it felt like the collective minds had been blown away was the day ChatGPT first came out. This felt like the ChatGPT moment for images — in a way that not even Stable Diffusion achieved, nor any other image model since. Certainly not in the same unified, joyful manner. Really, there were so many Ghibli images that the OpenAI GPUs melted.
There are so many ways to parse it. The first is the slight tragedy of it all. Miyazaki, the founder of Studio Ghibli, has expressly stated his deep dislike of AI image generation and the subversion of the artist, calling it “an insult to life itself.” While it certainly felt like each individual generation was a testament of love for such beautiful design — and in no way mean-spirited — it’s hard not to feel a bit wistful at another chip off the altar of the human soul. Nabeel put it simply:
Many people misunderstood Nabeel’s intent as being critical of the release, or even of the act of creating all those Ghibli images. I don’t think that’s what he meant. Maybe the word “sloppified” was not quite right — it implies the images themselves were slop, which I don’t think is true. He further clarified:
The whole thread had a lot of interesting comments back and forth. I bring it up here because it’s a great example of what’s to come. The killer application of AI has continued to be on the creative side, and we’re going to contend with these issues over and over again. Samples of writing and of drawings and of videos are all already online; they’ve already been consumed as training data. It’s hard to go back in time and remove them from existence. I’m honestly not even sure it’s possible for the AI labs to do so. There’s simply too much data. Labs can try to post-train, align, filter, and copyright-guard, but candidly, I don’t think it will ever be perfect.
And I think you can feel both ways. I felt excited; this launch was incredible, and there are so many fascinating downstream implications. Not just for fun images, but also professionally — imagine the ease of generating frontends now, paired with AI-assisted coding! IMO, in the same way that AI may help non-coders code (but likely won’t replace top engineers anytime soon), I feel the same about art. It allows non-artists to experience the beauty of creative power. As an art lover with 0 artistic talent myself, it feels nothing short of delightful.
But I think it’s okay that it feels a bit sad, too.
The Ghibli images are a strange sort of magic. They hit that emotional register that Ghibli films are so good at: a feeling of tenderness and yearning for something ephemeral. That they can now be generated by typing a few words into a box makes you wonder what else, what deeper parts of human expression, might soon be recreated this way.
Because that tenderness in the originals came from somewhere. From someone. The slow care of an artist’s hand, the years spent shaping a style, the emotional labor of translating a vision into something that speaks to others. When anything can be made by anyone in seconds, it’s not that the results stop being special, but their meaning does change. Mass production dilutes meaning.
That is what is at stake. The infinite availability of a style makes it feel less rare, and less rare often becomes less cherished, and thus creation is devalued. Not because the art is any less beautiful, but because the act of making it becomes harder to justify. Why spend years developing a style when it can be conjured in a moment? Why labor at all?
And that’s the real question, I think. We always thought what made us human was the capacity to dream, to create, to tell stories that made others feel. What happens, now that is replicable?
Everything must change.
When I was 18, I hopped on a plane to come to America by myself. I left behind everything and everyone I knew and loved. And I left behind some dark and difficult things, too. I chose to board that plane for a promise of a better life; I had hope, but no guarantees. I feel much the same way now. It is impossible, for me at least, not to feel longing and even mourning for a world that I love and that, if we face the obvious, we understand we will leave behind. The world that comes holds much promise – of diseases cured, of breakthroughs in food production, of many forms of abundance discovered as we now exponentially multiply our intelligence. And so I board the plane, and work so that others board it too.
But it’s okay, I think, to miss what we leave behind.
