Melvine's AI Analysis # 53 - 🚀 - Artists Collective creates Art using an AI model that reads their mind
Melvine Manchau
Senior Strategy & Technology Executive | AI & Digital Transformation Leader | Former Salesforce Director | Driving Growth & Innovation in Financial Services | C-Suite Advisor | Product & Program Leadership
May 4, 2025
This week, I had the incredible opportunity to attend a presentation by the Obvious Collective, a group of French artists and scientists based in Paris. The Sorbonne alumni club of New York organized the event. The collective presented its approach to visual art using AI. They started more than a decade ago and already have incredible success with their creations. You may remember their portrait of Mr Belamy, an AI-created portrait of an English lord, sold for half a million dollars at Christie’s.
They have multiplied the experiments by feeding classic and international art to their machine. They fine-tune their models by feeding particular sets of data to ensure the model takes the direction they are aiming at. The creations are then painted or handcrafted, like this extraordinary AI-created African mask crafted by artisans in Ghana.
What blew our minds was when they presented their latest work.
This new artwork is based on a known corpus of research around the translation of images visualized by the brain and recognized by the model using an MRI.
When a subject looks at a picture, some very particular parts of their brain light up and activate. The model is then trained to match the brain signal with the associated picture. It’s classic supervised learning. After a while, the model can recreate the pictures that the subjects thought about with little difficulty. They called this the Weak imagination process.
Where they pushed the envelope is in devising the strong imagination process. We are now moving to the unsupervised versions of the model. The subject is no longer going to look at a picture. They are going to “imagine” a picture. The distinction is crucial here. The subject is not just using its visual sense, which has a very physical cognitive path in the brain. The subject is now using a different part of the brain to “create” a picture from the mind. The algorithm now decodes the signal from the brain and then generates an image. The team worked with automatic writing techniques from the surrealist to “prompt” their brain. They read and memorized those writings as poems and imagined the pictures coming to mind based on those poems. The same way you and I imagine landscapes, decor, and characters while reading a novel or a poem.
The results are mind-blowing. The machine now creates a picture based on an imagined image. And that’s where the art comes in. We are not just seeing the best replica but the artist's mind itself. The team can now play with the model parameters or the artist's imagination. The work is grueling. It implies hours of lying down in uncomfortable positions in fMRI machines, very late at night, when the team can borrow a couple of hours, here and there of machine time. Esthetically, the results are superb, but the approach is what makes these creations formidable. And imagine all the possibilities that this capability could bring to the world. Entire sets of pictures, movies, plucked from our minds and regenerated in reality. The field of applications is exponential.
After the presentation, I discussed scalability with one of the artists. He mentioned that the experience is almost impossible to scale due to the heft of the process. But there is hope: smaller, still costly, and heavy brain scan machines are being developed and deployed. I asked about EKG helmets, and he demurred that the signals from these devices were too weak to be interpreted correctly by the models.
So, we are a far cry from the retail customer experience on these devices. But the century is young, and if these roaring twenties are teaching us anything, it is that there is certainly more to come in terms of technological prowess.
I also had an incredible conversation with 2 scientists about the future of molecular biology. But this is for another post.
I want to thank the team that put this extraordinary event together. The obvious collective is now raising funds for their next big thing. If you are a mécène, a patron of the arts and sciences, reach out to them. They will fill you in on this still-confidential project.
The future of art is bright. From the copyright fears and challenges of the phase we are in, we will surely move in the future into an explosion of creations, really imagined, thought, and crafted by the artists and scientists that are pushing us further into a formidable future.
By Melvine Manchau, Digital & Business Strategy at Broadwalk and Tamarly
https://melvinmanchau.medium.com/
