Programme timeline
10/09/2025
Short introduction
Zuzana Husárová | Karel Piorecký17:00-17:10
Connectivity as Device. Contemporary Poetry and LLMs
17:10-17:30
The most prominent program currently operating with ANN are the GPT large language models (LLMs) developed by OpenAI. They are capable of generating outputs that align with the statistical calculations of the training data and bear a resemblance to the input data. The underlying model maps the respective connection strengths between the “neurons” and the necessary calculations are executed in parallel within the network of neuron layers. The process itself remains opaque; only the output can be observed, not the way in which connectivity – i.e., text composition calculated by using AI – takes place. As key example, one of the four volumes published by 0x0a, a collective for conceptual digital literature consisting of Hannes Bajohr and Gregor Weichbrodt, will be discussed: Monika Rinck. Poetisch denken 1. GPT-2 was trained with all published poetry by Rinck that Christian Metz examined in his study Poetisch Denken. Die Lyrik der Gegenwart. However, the programming itself remains undisclosed; it was merely stated that the training was interrupted before the poems became “too good,” as the objective was to accentuate the output as a concept itself. While the data set fed into the ANN is known here – in contrast to all versions from GPT-3 onward – the precise functionality of the LLM remains opaque. The paper will use this “secondary” poetry publication to discuss the similarities and/or divergences between poetry-specific methods of “connectivity” and those of LLMs. The starting point are approaches in literary genre theory that consider poems as connectionist or network-like linguistic entities. This includes Roman Jakobsen’s axis model and equivalence principle, Jürgen Link’s concept of “excess structuring,” isotopes as “semantic networks,” and Espen Aarseth’s concepts of ergodicity and multicursal movements, developed within hypertext theory.
Adversarial Noises
17:50-18:10
the lecture performance „adversarial noises“ will demonstrate the creative misuse of speech and text generation by neural networks and generative artificial intelligence for poetry and especially sound poetry as well as adversarial attacks on speech recognition and training data for large neural networks in order to disrupt and subvert their function.
Discussion
18:10-19:00
11/09/2025
A new archivist has arrived in town... Generative AI as the new curator.
16:00-16:15
The current discussion on the use of artificial intelligence (AI) in the humanities and creative professions mostly focuses on the impacts in the sphere of creative authorship – generating images, texts or melodies using freely available AI applications. However, artificial intelligence has also recently been given increased attention in the field of curatorship, focusing on the preservation, organization and exhibition of works of art and cultural artifacts. In the presentation, I will focus on those examples of the use of AI in curating that go beyond the traditional archival work towards the concept of curatorship as a creative profession, in which the authorial gesture of the curator is significantly reflected, offering a new perspective not only on the exhibited works and the relationships between them, but also on the formal aspects of the process of creating an exhibition and on the exhibition as a medium mediating artworks. Based on examples of exhibitions curated using AI, the presentation will seek an answer to the question of what forms and functions AI acquires in the creative process of curating. The use of AI in curatorial practice will be understood as contributions to the development of techno-diversity through the creative appropriation of AI technology and experimental exploration of its potential.
Digital Humanities and LLM: A Perfect Match?
16:15-16:30
Will large language models help make digital humanities mainstream and combine qualitative and quantitative research in literary theory? And how many pubs did Švejk actually visit?
How LLM Created Work instead of Saving it
16:30-16:45
The use of LLMs in education is most often associated with making students' work easier, sometimes even cheating. In the Digital Arts studio at the Academy of Fine Arts and Design, we used a model for generating poetry as part of the assignment. In doing so, we have turned the debate about the use of LLM on its head: How can LLM generate new creative opportunities for us instead of killing existing ones?
Discussion
16:45-17:15
Coffee break
17:15-17:30
Presentation of Students' Works from Digital Arts AFAD
Marty Müller | Natália Barnová | Jakub Cingeľ | Kristína Matisová | Olívia Jánošíková | Marco Fíšan | Vira Chernomorets17:30-18:30
PoetPet
Marty Müller | Natália Barnová | Jakub Cingeľ17:30-17:45
PoetPet is a playful, educational video game where players create poetry by jumping on platforms. After designing their own poet-animal avatar, players compose poems by physically following a given rhyme scheme—jumping on a dance mat to reveal verses. Powered by the EduPo generative AI model, the game produces original poetic lines tailored to each avatar’s unique characteristics. With an aesthetic inspired by Tamagotchi and movement-based interaction, PoetPet invites players to feel, catch, and quite literally jump into the rhythm of language.
Magické doupě
Kristína Matisová | Olívia Jánošíková | Marco Fíšan17:45-18:00
Magické doupě is a digital video game that throws players into a fast-paced language battle with a human-like entity in a virtual lair. Part poetic match, part magical showdown, it’s a duel where clever verses score points—and the wittier wordsmith wins. Supported by the EduPo AI model, the wizard’s mouth conjures real-time rhymes, daring players to keep up, strike back, and out-poet their digital rival. The action spills beyond the screen into a physically immersive setup that mirrors the room of the virtual duel. Sharpen your tongue, steady your stance - Magické doupě dares you to outwit your enchanted opponent and break his will to win.
Your Tongue Is Not My Language
Vira Chernomorets18:00-18:15
Your Tongue Is Not My Language is an immersive poetic installation that employs the generative AI model EduPo to guide viewers into a multilingual experience of reading and understanding language from a non-native perspective. It invites visitors into a realm where letters blend, verses blur, and precise meaning dissolves into a tapestry of emotional ambiguity. The work evokes the experience of foreigners immersed in linguistic disorientation—where words resemble images, and meaning is sought not through language itself, but through emotion, intuition, and affective resonance.
Discussion
18:15-18:30
GenGen Exhibition Opening
Marty Müller | Natália Barnová | Jakub Cingeľ | Kristína Matisová | Olívia Jánošíková | Marco Fíšan | Vira Chernomorets18:30-20:00
About the Conference
The development of so-called artificial intelligence, the refinement of its outputs, and the expansion of its possible applications are all progressing at a breathtaking pace. Literature and visual art are by no means exempt from this trend—on the contrary, these traditional domains of human creativity were among the first in which contemporary AI was applied and tested. What forms does the relationship between art, literature, and AI take today? What is the nature of creativity that combines human and machine capabilities? And how is our perception of literary and visual works changing in the context of omnipresent synthetic images and texts? These are just some of the questions that will be addressed in the presentations and discussions featuring internationally recognized experts as well as specialists from the Czech Republic and Slovakia.
Conference on Generative AI in Art and Literature
10. – 11. 9. 2025
Praha, Kampus Hybernská - Didaktikon