International Semiotics Seminar in Paris

The Semiotics in Paris 2024-2025 seminar will continue the collective research begun in 2023-2024, focusing on theories of enunciation and the impact of passions in the semiotic territories opened up by artificial intelligence. The previous edition explored a range of topics, including robotics, immersive arts, methods for analysing data corpora, and generative artificial intelligence. This year, the focus will be on a specific theme among those already addressed: computational models of languages and translation problems (between human prompts and machine-revised prompts, between natural language prompts and textual descriptions of images), in particular generative artificial intelligence.

The aim is to continue thinking about the multiple enunciative instances involved in these intelligences: the databases used to learn generative models, the training and inference algorithms, the numerical representations called ‘embeddings’ which translate visual or verbal texts into lists of numbers, the codes which articulate these elements in a specific programming language, and finally the ‘prompts’, which serve as an entry point for the user. This approach will make it possible to focus on the many translations between different syntagmatic expressions, while addressing the complex issue of creativity. Indeed, enunciative instances emerge in texts produced by generative artificial intelligence with varying intensities of presence, combining in various ways. These combinations of human and non-human actions provide a better understanding of how creativity manifests itself in the discourses of generative artificial intelligence.

Creativity will be studied through different domains, such as visual arts, literature, argumentative logic and common sense. The aim will be to differentiate the various forms of creativity according to the universes in which it is deployed: visual arts, graphic design, advertising marketing, politics, strategy games, jurisprudence and decision-making.

In these fields, creativity is evaluated according to its own specific criteria, which vary according to the discursive genres in which computational tools are used. It manifests itself differently in the hard sciences and the humanities. Generative artificial intelligence thus penetrates all aspects of collective and individual life, operating in a variety of ways depending on the context. In each case, the enunciative instances differ in the process of recombining texts already written and their impact on users, thus influencing the production and ongoing reconfiguration of this new discursive economy. The case of generated images is particularly emblematic: each image is perceived more as the result of an exploration of a latent space of virtualities than as an individual product.
The object of study will not be limited to the semiotic analysis of texts co-produced with artificial intelligences such as ChatGPT, DALL-E, Midjourney or Stable Diffusion, but will extend to all socio-cultural practices involving generative artificial intelligence, whether in education, political debate, scientific research or artistic production. It will also examine the computer practices that implement and continually modify them. In this context, particular attention will be paid to the study and evaluation of what is termed the ‘meta-viewpoint’ produced within the operational chain of generative artificial intelligences, which can be interpreted as a form of self-monitoring attributed to the machine.

In parallel, other human attributes delegated to machine operations, such as perception, will be studied. Recent literature on the subject mentions concepts such as ‘perceptual topology’ or ‘archival perception’. Are these merely metaphors? The productions of Midjourney, DALL-E and Stable Diffusion show a certain management of spatiality in the composition, respecting perceptual affordances, particularly in the staging of actions and objects. But how can we conceive or describe a perception without a body? These questions (translation between human and machine languages, creativity, impact on different social domains, self-reflexivity of the machine, perception) will be addressed throughout the year from a theoretical, methodological and genealogical perspective, in order to understand the evolutions, ramifications and specialisations of generative intelligences.

Practical information

• Sessions will be held at the Maison Suger - 16-18 rue Suger, 75006 Paris
• Dates and times: every Wednesday from 1.45pm to 5pm

Scientific council

Juan Alonso Aldama, Pierluigi Basso (coordinateur), Denis Bertrand, Anne Beyaert-Geslin, Jean-François Bordron, Marion Colas-Blaise, Nicolas Couégnas, Ivan Darrault-Harris, Rossana De Angelis, Valeria De Luca, Maria Giulia Dondero, Veronica Estay-Stange, Jacques Fontanille, Didier Tsala-Effa