AI, art and neuroscience. A conversation with multidisciplinary art studio fuse*

Fuse* is a multidisciplinary art studio based in Modena, Italy, founded in 2007 by Luca Camellini and Mattia Carretti. In the last 15 years, the studio has been researching the expressive possibilities of emerging technologies, aiming to interpret the complexity of human, social and natural phenomena. Their work (primarily expressed through multimedia installation and performances) has been exhibited all over the world, often involving data sets, artificial intelligence, and collective creativity, and aims to explore the boundaries between different disciplines in pursuit of new connections between light, space, sound and movement. We had the pleasure of interviewing them about their latest projects.

Many of your works – such as Onirica (), in which you use data sets to give shape to dreams, but also Fragile, that exploits the scientific data provided by the Society for Neuroscience – start from scientific research, particularly linked to neuroscience. How does artistic experience help and strengthen scientific research?

Within our studio, the intersection between art and science has always been fundamental in terms of narrative, concept and audiovisual development. Often our inspirations are rooted in research or studies carried out by scientific realities, and this may derive in part from the fact that we come from heterogeneous paths of study and expertise, first of all the founders and directors of the study: Luca Camellini, who is a computer engineer, and Mattia Carretti, who is a graduate in chemistry.

In general, curiosity is an essential driver: we organize frequent internal sharing sessions regarding both innovations in new technologies and discoveries and issues that team members consider relevant to be shared with the rest of the study. Other times, however, we come across stories in which we immediately see great potential, as in the case of Onirica (): we learned about the Laboratory of Psychophysiology of Sleep of the University of Bologna through another project, and we were immediately fascinated by their Bank of Dreams. From the very beginning we understood the power of using this database as a basis for a story.

Luna Somnium, large scale installation, 2022

What is the power, and what is the need, to put artistic production at the service of science (and science at the service of art)?

Generally speaking, we feel that art and science reinforce each other: on the one hand, we perceive that when a work has a solid scientific basis and a well-defined message to communicate, it takes on a completely different meaning and power. On the other hand, conveying scientific messages – which are often complex – through artistic language makes them accessible to a wider audience, thanks to the emotional connection that is established between the work and the user. Often, theories or scientific discoveries fail to touch non-experts, remaining distant from common perception: explaining these issues through an emotional and engaging language makes them more accessible and easier to understand.

The theme of the night, of the dream, runs through the last two works – Onirica () and Luna Somnium. What fascinates you about the exploration of the "alternative reality” that takes place during the hours of sleep?

It’s always interesting when you point out these kinds of connections, especially when they’re highlighted by external eyes – our view is so subjective that sometimes we lose such connections. Both projects undoubtedly highlight an interest in the dimension of the dream: while Onirica () is placed in direct connection with the theme, for Luna Somnium the relationship emerged later, when we were looking for a narrative that could integrate with the installation in order to add a level of reading to the audiovisual experience.

One of the most fascinating aspects of the dreamlike dimension is the ability to perceive reality in a unique and unprecedented way, offering the opportunity to live experiences that are often extraordinary and difficult to replicate in everyday reality. This concept is rooted in  our vision and the reason why we do what we do: to give the public the opportunity to live a unique and profound experience, being able to suggest a new interpretation of themes which are more or less familiar, but which are often observed superficially, on “autopilot”. The dream serves as a starting point to explore narratives which are out of the ordinary, stories that we all share as human beings but that lead us to discover reality with more vivid eyes.

Onirica (), Ugo Carmeni Studio, 2023

In immersive installations such as Trust, there is a fundamental desire to use new technologies to deliver an experience that arouses "empathy". Yet, is it possible to "measure" empathy, evaluating the impact of experience?

So far, we have never had the opportunity to organize activities of objective analysis of the effectiveness of our installations, mainly because of the difficulties in defining moments of comparison or in formulating questionnaires for visitors. Our main channels of communication with the public are social media, where we can perceive the reception of a project through the level of interaction with specific content. Also, through this channel we frequently receive messages of interest or feedback on the works, installations and performances that we share: when a work affects the audience and is received in a positive way, it is perceived quite clearly. In addition, although it is not always possible, we like to be physically present at vernissages to collect direct feedback and to see with our eyes how visitors interact with installations.

In your experience, what are the criteria for measuring the effectiveness of the installation, when it comes to increasing the empathic response of the observer?

Although we have not yet implemented moments of feedback from the public, for each project and exhibition we arrange internal sharing sessions with the fuse* team and share more specific forms with external collaborators who have supported us in the process: these moments are always very useful and offer us the opportunity to see the whole project with eyes that are a little more detached, deepening what worked and what didn’t. The most important thing, however, is to be able to have feedback on the significance and impact of a specific installation.  Although they are often subjective elements, we always notice a certain consistency between internal feedback from employees, and public perception.

Artificial Botany, A/V installation, 2019/ongoing

You already make use of machine learning in Artificial Botany, but also in Onirica () generative artificial intelligence is a core point. Do you share the widespread enthusiasm for the potential of AI in the production of images and in the collaboration between the human and the artificial, or do you find any critical issue?

In Artificial Botany, the goal was not quite to explore the final result of the generation obtained by the GAN (Generative Adversarial Network), the machine learning system at the base of the generations, but rather the process itself: we try to explore that intermediate dimension that emphasizes the metamorphosis of forms and their continuous fluidity. In this case, the technology that is used creates an interesting contrast with the organic world, reinterpreting the original scientific illustrations with new eyes.

In the case of Onirica (), AI played a central role in the development of the entire project because the visual part is the result of a text-to-image diffusion model: a generative system that translates text indications into images. We always want to emphasize that AI does not, of course, have autonomous decision-making or creative abilities; it is the user who must guide the generation, defining the aesthetics and visual choices.

By working so closely with this tool we experienced some of its limitations, the most glaring ones relating to the generated subjects and the interpretative prejudices inherent in the system. For example, all human figures presented the same characteristics and the same level of aesthetic perfection, as they were taken from advertising or (especially for female figures) from 19th century paintings. Being able to differentiate the subjects in terms of age, build and cultural heritage turned out to be particularly complex and that’s why we realized that we also have to learn to communicate with the system, using a language that could be understood by the model and that could give the result we had in mind. It is precisely these limitations that we want to make visible through Onirica (): like many of our works, also in this case we wanted to experiment with a new technology and we are happy to have had the opportunity to deepen this world in such a direct way, bringing to light the limits of an artificial system that is instead often considered to be already capable, already intelligent.

Laura Cocciolillo
Scroll to Top