6 days Ago By [email protected] (Shubham Verma)
Updated April 1st 2025, 17:16 IST It is easy to overlook these nuances of using an AI assistant, but at scale, AI-driven trends could have long-term ecological consequences. Studio Ghibli was famous but niche. ChatGPT turned it into a global phenomenon for good and bad.
OpenAI’s bot sent the internet into a frenzy after it gained the capability to generate AI images as part of GPT-4o integration. While the new capability allows ChatGPT to offer several different art styles, Studio Ghibli stood out. So much so that it went viral, crashing the company’s servers, and stressing the supporting hardware.
Users quickly found out that by just uploading any photo and giving a command, they could get an adorable rendition of that photo in Hayao Miyazaki’s Studio Ghibli style on the screen. All this is cool, but it may be just another fad until the internet finds another cause to fixate on. And as much fun as turning favourite family photos or just any random snap into Studio Ghibli is, its environmental cost may persist beyond the viral moment.
It is easy to overlook these nuances of using an AI assistant, but at scale, AI-driven trends could have long-term ecological consequences. And they could be irreparable — at least that is what environment experts believe. A single AI-generated image, including that with Studio Ghibli style , consumes anywhere between 0.
01-0.1kWh, generating enough heat that requires about a teaspoon of water for cooling. That may be insignificant in isolation, but imagine millions of people generating more than one AI photo every minute.
The higher the load is on data centres, the more water is required to cool them. OpenAI CEO Sam Altman on Tuesday announced that the company added 1 million users in a single hour , largely because they wanted to jump on the Studio Ghibli art bandwagon. With the existing rate limits of 3 AI images per day, multiplied by 1 million, the total count of AI-generated images stands at 3 million per day.
But this is just for the users who joined ChatGPT in one hour. Factoring in the existing users will take up that number even higher. It would take no less than 7,500 litres of water, equivalent to the daily water use of around 100 people by global standards.
Now, keep the water aside and talk about the carbon footprint. ChatGPT has mentioned that generating a single AI image can emit between 5 and 50 grams of carbon dioxide. An estimated 3 million images a day release roughly 1.
35 metric tonnes of carbon dioxide. It is equivalent to driving a petrol car for about 5,000km, charging over 162,000 smartphones simultaneously, streaming over 110,000 hours of Netflix, and even flying from New York to London and back. As long as the data centres are powered by renewable energy, the net emissions will become zero.
However, OpenAI has not disclosed if its data centres are using any energy that replenishes itself. So, it is a long shot to think that the net carbon emissions from generating Studio Ghibli images daily are zero — at least currently. There is no way AI trends will abruptly stop, but, as Altman emphasised , “y’all need to chill please.
” And since that is almost unlikely, a switch to green energy for running data centres could contribute to curbing the long-term impacts. Altman said OpenAI would scale up the capacity of the model but did not clarify if that would strain the environment even more. The Ghibli-fication of photos also raises ethical and moral questions, reminding users that creating a true masterpiece requires skill, not just a click.
Published April 1st 2025, 17:16 IST.
