Radical Desserts is a media art project composed of 150 video clips and a website designed in the style of a “dessert blog.”
The idea was born during a Sicilian pastry course I attended in Rome some time ago. That’s where I met Jamila Alaoui, a young pastry assistant who, thanks to her skill and passion, seemed to know even more than the instructor himself.
It was largely thanks to her contribution that I managed to put together the demanding initial list of desserts. Once the list was complete, I used artificial intelligence to generate the corresponding images—without yet having a clear idea of how to use them.
About a month later, I came across a video documenting the presentation day of the 61st Venice Art Biennale, curated by Koyo Kouoh, who had unexpectedly passed away about two weeks before the event.
What struck me most was the ending of the video, where a young curator read a few moving lines that Koyo Kouoh had written shortly before her death:
We need to laugh.
We need to dance.
We need the radicality of joy.
Inspired by those words, I decided to pair the dessert images with short texts written in the form of tercets. I began asking the AI questions such as: “What kind of language would a dessert use if it could speak?”
After several experiments, I refined the prompts by drawing inspiration from Dadaist poetry and comic book language. The AI then generated a series of tercets—each paired with a dessert and inspired by Koyo Kouoh’s words.
At that point, I combined the texts and images, first experimenting with synthetic voices and then—with a little help from a WhatsApp message—the pop-up windows typical of messaging apps.
Finally, using specialized software, I produced the video clips, designed the graphics, and developed the project’s website.
Dario Quaranta Neropop (media artist and web editor – lives and works in Rome)
Jamila Alaoui (pastry assistant and cake designer – lives and works in Rome)
Technical Details
Radical Desserts is a media art project composed of a website and 150 video clips in 4K resolution.
The video clips are divided into three collections of 50 clips each.
The project was created using two different AI platforms—one for text generation and another for image generation—along with a range of specialized software for video editing, graphic design, and web development.
