The AI Studio "Ghibli" Mode Storm: All You Need to Know

From Abhivardhan, our Chairperson

Since 'Ghibli Storm' has become so popular, I know there are firms claiming 'ethics' and 'regulatory' expertise, so much so that they put their team photo to 'ghiblification' via OpenAI's tools, not caring about their privacy.

Anyways, here's my take on this 'storm' which even enchanted Government of India and the French Government. And I am not going to discuss copyright law. It's boring:

1️⃣ No, I’m not interested in this Ghibli mode by OpenAI. Don’t use my pics nor would I waste my time doing it.

2️⃣ The fusion of AI to make art is great. I support creators like Prateek Arora and others who genuinely use GenAI outputs, and then cultivate their own AI workflows, in a way to shape up their creative pursuits. However, this 'ghibli' thing was among the most exploitative visual data crunching campaigns by OpenAI. They are not a product company, since they are a research firm and that’s it. They are not like Perplexity. So we don’t know how their policies imply anything.

3️⃣ People fell into the cuteness of these trendy images and willingly shared their own family images consensually to a company like OpenAI. That is so lazy. Now, this doesn’t mean we should not use AI tools in artistry, but it feels very lazy and performatively fake to me.

4️⃣ No, I don’t think any so-called graphics designer jobs are gonna be over. This Ghibli mode only strengthens the need to have smart designers who use the right AI tools. If you think you need a lazy solution for everything, then you are being fed something comforting which is not creating anything new. And design needs are human. You can create workflows which make you productive, for say 3D printing and AI, and that's fine. But the Ghibli thing isn't making you productive.

5️⃣ Or should I now assume that being innovative is to assume a generic way of converting real images into generative template-like outputs - then we are not achieving anything in computer vision + generative AI. This whole trickery by OpenAI doesn’t achieve anything specific on artistic workflows and their fusion with AI nor it tells the avenues around AI + art research.

6️⃣ Now, new models of copyright and licensing are needed but a good case can be made against OpenAI on a goodwill issue rather than the traditional way we see copyright. Or maybe we can inspire from a US-based case law where no such “Ghibli”-style images are copyrighted by the one who prompts it to get. But that also means you can’t demonetise creators. Will that happen? I don’t know.

OpenAI is an exploitative company and can’t be trusted. Anyways, these are cute images, so if you are enjoying them, that's your fun. I don't mind. But that's just an act of desperation by a company that can neither be a proper research firm, nor a product company. I don't know why stakeholders in Delhi are so much wooed by OpenAI like that.

Now data sovereignty doesn't matter? What do you think?

I have discussed that AI Studio "Ghibli" style (and aesthetic) won't take away the jobs of graphic designers. I have given my technology law perspective anyways (check out: https://www.linkedin.com/posts/abhivardhan_since-ghibli-storm-has-become-so-popular-activity-7312277487529406464-cSwq?utm_source=share&utm_medium=member_desktop&rcm=ACoAAB3bBjMB8v_lDnQkP5VGc3DVHLlJvJELCGU)

But as someone who has written 450+ poems in Hindi and English combined (yes, yet to publish), I love art, cinema and music to its core. I love the jingles of Ilayaraaja, the symphonies of Brian Tyler, and even the classics of Ennio Morricone, Kenji Kawai and Ajay-Atul.

Here is an interesting arts explainer video by Prachi P. (prachipop.studio on Instagram) why OpenAI has done something even simplified, and mediocre.

1️⃣ I use some AI video and photo generator tools, and have realised that these tools regurgigate the same, old content.

2️⃣ Now, people will claim how does it not make GenAI innovative? Let me explain you with a story.

There is a character in Mahabharata called Ashwatthama. I wrote about Ashwatthama a poem in Hindi, a decade ago, and my perspective is modernist, but what Nag Ashwin showed in a movie like Kalki 2898 around Amitabh Bachchan's Ashwatthama does not need to be the same perspective that I had right?

3️⃣ Art is about access of ideas with newer formats. The problem with OpenAI's deliverable is that it generalises the soulful sadness, happiness and anti-industrial visual silence that folks like Miyazaki at Studio Ghibli produced.

Flooding the internet with surface-level and easily replicable automatic aesthetic style is not artistic, is empty by heart but it looks pretty for sure.

4️⃣ I like the example of Raja Ravi Verma explained by Prachi in this video. What Raja Ravi Verma did was that his printing press of his oil paintings made his works accessible to masses. His interpretation of mythology, beauty and realism was democratised to so many people.

5️⃣ Also on those people making Ghibli-knockoffs of Denis Villeneuve's Dune: Part Two: it's quite lazy of you to use Sora Remix, then Ghibli-style, then Kling, and then CapCut 🫠 😆 - you are just trying to create cheap knock-offs.

Again, it's shiny and appealing, but it doesn't create new artistic workflows embedded with AI, nor would help human stakeholders in the animation and cinema industry, let alone gaming.

It's useless to regurgitate our past using "Ghibli-style" and then keep digging the hole. It's boring. Go, learn some artistic style, musical notes, some raaga, or some AI ethics ideas published by Indic Pacific Legal Research LLP at https://indopacific.app/store 😂 .

Stay curious. Trust, but verify.