Translation with the GPT LLM. A case study of the "King UBU" by Alfred Jarry
Submission Type
Paper
Start Date/Time (EDT)
20-7-2024 10:30 AM
End Date/Time (EDT)
20-7-2024 11:30 AM
Location
Algorithms & Imaginaries
Abstract
The translation, and publication of a book titled "UBU GPT" based on the Alfred Jarry's masterpiece is a mockery of the process and of the book edition of the translation. A mockery, let us add, rooted in the style of the "Ubu King's" author. The Polish edition includes a translation, a plot summary and a re-writing of the drama from the summary, as well as a transition of the drama into a contemporary novel using youth language (including obscenities); it ends with an essay on translation and application of LLMs. The main text was provided by the GPT-4 model; the prompts were designed by Jan K. Argasiński (UBU Lab), and the publishing and marketing was handled by Piotr Marecki (Ubu Lab).
Our paper is about the translation of Jarry done by the GPT; as well as LMM translation into Polish. We also report on introducing the finished product to the market.
Translation/paraphrase using GPT (or any transformer) is interesting process by itself. GPT first “understands” the meaning of the English text, with a goal in maintaining its essence. Then, leveraging available training data, applies grammatical rules and Polish vocabulary to rephrase the content. This involves algorithms for including context, cultural nuances, and idioms to generate appropriate outputs. The whole process leads to interesting and meaningful errors and quirks that we in our paper call “mockeries”.
keywords: artificial inteligence Ubu King GPT LLM
References:
Jarry, A., & Tiesset, L. (2011). Ubu roi. Bouquineo.
Wolfram, S. (2023). What Is ChatGPT Doing... and Why Does It Work?. Stephen Wolfram.
Wu, Y., & Hu, G. (2023). Exploring prompt engineering with GPT language models for document-level machine translation: Insights and findings. In Proceedings of the Eighth Conference on Machine Translation.
Recommended Citation
Marecki, Piotr and Argasiński, Jan K., "Translation with the GPT LLM. A case study of the "King UBU" by Alfred Jarry" (2024). ELO (un)linked 2024. 17.
https://stars.library.ucf.edu/elo2024/algorithmsandimaginaries/schedule/17
Translation with the GPT LLM. A case study of the "King UBU" by Alfred Jarry
Algorithms & Imaginaries
The translation, and publication of a book titled "UBU GPT" based on the Alfred Jarry's masterpiece is a mockery of the process and of the book edition of the translation. A mockery, let us add, rooted in the style of the "Ubu King's" author. The Polish edition includes a translation, a plot summary and a re-writing of the drama from the summary, as well as a transition of the drama into a contemporary novel using youth language (including obscenities); it ends with an essay on translation and application of LLMs. The main text was provided by the GPT-4 model; the prompts were designed by Jan K. Argasiński (UBU Lab), and the publishing and marketing was handled by Piotr Marecki (Ubu Lab).
Our paper is about the translation of Jarry done by the GPT; as well as LMM translation into Polish. We also report on introducing the finished product to the market.
Translation/paraphrase using GPT (or any transformer) is interesting process by itself. GPT first “understands” the meaning of the English text, with a goal in maintaining its essence. Then, leveraging available training data, applies grammatical rules and Polish vocabulary to rephrase the content. This involves algorithms for including context, cultural nuances, and idioms to generate appropriate outputs. The whole process leads to interesting and meaningful errors and quirks that we in our paper call “mockeries”.
keywords: artificial inteligence Ubu King GPT LLM
References:
Jarry, A., & Tiesset, L. (2011). Ubu roi. Bouquineo.
Wolfram, S. (2023). What Is ChatGPT Doing... and Why Does It Work?. Stephen Wolfram.
Wu, Y., & Hu, G. (2023). Exploring prompt engineering with GPT language models for document-level machine translation: Insights and findings. In Proceedings of the Eighth Conference on Machine Translation.
Bio
Dr Piotr Marecki - associate professor at the Jagiellonian University. Writer, digital media artist, publisher, translator and digital culture scholar. The head of Ha!art Publishing House (ha.art.pl) and of UBU lab (ubulab.edu.pl). In 2013-14 he did a postdoc at MIT at the Trope Tank lab. Based in Kraków, Poland.
Jan K. Argasiński, PhD - Media theorist and philosopher. Cyberpunk. Occasionally demoscener. Works at the Department of Games Technology of the Faculty of Physics, Astronomy and Applied Computer Science, Jagiellonian University. In his day job he works on the processing of emotions in computer systems (affective computing) and on virtual and augmented reality projects.