Programme > Courses

Generative Models for Imaging – Andrea Asperti

This 4-hour course offers Ph.D. students an introduction to cutting-edge generative models in imaging, with a special focus on the rapidly evolving field of diffusion models. From photorealistic image synthesis to super-resolution, inpainting, and style transfer, generative models are reshaping how we create and manipulate visual content.

We'll begin with a quick tour of the major families of generative models, before diving deep into diffusion models—the driving force behind today's most powerful image generation systems. You’ll explore how conditioning techniques can be used to steer these models, enabling the creation of images with targeted structures, textures, or styles.

The course also critically examines the current challenges and limitations of generative approaches, especially in tasks involving creative control and style consistency.

By the end of the session, you’ll gain a clear, practical understanding of diffusion models: how they work, where they excel, where they struggle, and how to apply them effectively in creative projects.

AI and Visual Culture: A Theory of Latent Spaces – Antonio Somaini

A theory of images and visual culture, today, needs a theory of latent spaces. In a historical phase in which images are more and more generated, modified, circulated, seen and described by or with the help of different kinds of AI models, we need to understand the crucial role played by an abstract, mathematical construct whose cultural and political implications could hardly be overestimated. Latent spaces play a crucial role in generative AI models, from GANs to the recent diffusion models. They also play a central role in the contemporary artistic practices that engage critically with AI, responding to its increasing presence in every aspect of culture, society, politics and economics. For a few years now, artists have developed different strategies to explore or modify the existing, dominant latent spaces, or to produce their own alternative, antagonist, counter-hegemonic ones. Considered together, these different strategies show the awareness with which the field of contemporary art is tackling the presence and the agency of this hidden layer of mathematical abstraction that participates in the shaping of cultural and political imaginaries.

Deepfakes and cultural industries: a legal perspective - Celia Zolynksi

 As generative AI becomes more prevalent in cultural industries, there is an ongoing discussion to establish a framework for these applications, with the aim of considering the interests of various stakeholders, including artists and the public. In the year following the implementation of the Regulation on Artificial Intelligence, this course will undertake a comprehensive examination of the legal framework applicable in this context, the questions currently being posed, and the potential avenues of development that should be considered in order to take account of the specific nature of this sector.

The new aesthetic effects of generative AI art - Jim Gabaret

AI-generated texts, music and images are often criticized for their flatness, their kitsch or the many sexist, racist, classist or youthist clichés they convey. Beyond the desire to reflect on the irreplaceability of Man in the face of the socio-economic and political risks posed by AI, a lot of critics defend that there is a profound lack of intentionality, corporeality, sensitivity, being-in-the-world and access to artistic meaning in these content-generating machines. But understanding the mode of existence of these new technologies allows us to move beyond the anxiety attached to the “copy paradigm”, to observe the creativity of AI art itself. Trained on large bodies of human culture, it offers “re-representations” of our social imaginaries, that reveal certain unnoticed aspects of our world of representations, in a way similar to the revelations we expect from artists. But we need "distancing devices" to appreciate their aesthetic effects beyond the fear of illusionism: it is by looking at them from their artificial origin that these works are most valuable.

Exploring the Convergence of AI and Virtual Reality in Art – Borja Jaume Perez

This course examines the intersection of artificial intelligence and virtual reality in contemporary artistic practice. Participants will learn about AI-driven creativity, the role of VR in immersive art, and the impact of these technologies on artistic expression. Through case studies, interactive exercises, and creative experimentation, we will analyze how AI can enhance VR environments, generate artistic content, and expand creative possibilities. The course also encourages interdisciplinary thinking, reflecting on the role of AI in redefining the role of the artist in the digital age. Upon completion, participants will understand how AI and VR may shape some of the future of artistic experiences.

NB: participants need to install Unity 3D on their work computer for this workshop. We will provide a download link with instructions.

Generative AI and Natural Language Processing for Creativity - Serena Villata

In this lecture, we will explore the intersection of Generative AI and Natural Language Processing (NLP) as tools for fostering creativity across diverse domains. I will cover the fundamental principles of generative models, with a specific focus on (large) language models like Mistral and GPT, and their application in generating novel text. The goal will be to discuss how current NLP techniques can enhance human creativity, support artistic expression, and transform industries such as literature, marketing, and entertainment. The course also examines the ethical considerations and challenges of leveraging AI for creative endeavors.

AI for Creativity in Psychiatry: Can AI help us understand mental disorders and improve our mental well-being? Tuomas Vesterinen

Researchers hope that artificial intelligence (AI) can provide novel methods to discover the causes of mental disorders and improve mental healthcare. However, the employment of AI in psychiatry raises unique ethical and philosophical questions due to the field’s wide scope and the special nature of mental disorders. Psychiatry holds social power, deals with human behavior and experience, and the nature of mental disorders is both complex and controversial. On the other hand, AI itself raises questions about the nature of creativity, agency and responsibility, among others. These lectures will explore whether, and how, AI could enhance scientific creativity to address these challenges, as well as the larger social and ethical implications that may arise.

Can AI Ever Be As Creative As Humans? Mark Keane

In these lectures, I consider how current GenAI claims about creativity fail to equal human creative acts in science and the arts.  Though some recent outputs from GenAI systems seem impressive (eg from LLMs), there is an interpretational and informational context to these demonstrations that is typically overlooked.  I argue that fundamentally there is a human context to the Creative Act that is not paralleled by current AI systems, that architecturally these systems are limited representationally, restricted in their world knowledge and poor at self-directed goal definition and problem identification. These differences between humans and machines raise questions about whether GenAI will ever equal the products of the human mind (though it may, of course, end up doing something that is quite different and also interesting).